Sofology Limited & Anor v The Commissioners for HMRC

Neutral Citation: [2022] UKFTT 00153 (TC)

Case Number: TC08480

FIRST-TIER TRIBUNAL

TAX CHAMBER

Taylor House, 88 Rosebery Avenue, London EC1R 4QU

Appeal reference: TC/2019/04308; TC/2021/00353

VALUE ADDED TAX - deductibility of input tax on the costs of advertising on Google – each Appellant made taxable supplies of sofas and exempt supplies of insurance intermediary services – held that the costs of the advertising had a direct and immediate link with the taxable supplies of sofas and therefore were not overheads – in addition, the costs did not have a direct and immediate link with the exempt supplies of insurance intermediary services – instead, the link between the costs and the exempt supplies was indirect – appeals upheld

Heard on: 7, 8, 9, 10, 14, 16 and 17 March 2022

Judgment date: 29 April 2022

Before

TRIBUNAL JUDGE BEARE

MS GILL HUNTER

Between

SOFOLOGY LIMITED

and

DFS FURNITURE COMPANY LIMITED

Appellants

and

THE COMMISSIONERS FOR HER MAJESTY’S REVENUE AND CUSTOMS

Respondents

Representation:

For the Appellant: Ms Valentina Sloane QC, of counsel, instructed by Deloitte LLP

For the Respondents: Ms Natasha Barnes and Mr Paul Reynolds, of counsel, instructed by the General Counsel and Solicitor to HM Revenue and Customs

DECISION

“In so far as the goods and services are used for the purposes of the taxed transactions of a taxable person, the taxable person shall be entitled, in the Member State in which he carries out these transactions, to deduct the following from the VAT which he is liable to pay:

the VAT due or paid in that Member State in respect of supplies to him of goods or services, carried out or to be carried out by another taxable person;...”

17. The principle of attribution is then set out in Article 173 of the Directive, which provides as follows:

“1. In the case of goods or services used by a taxable person both for transactions in respect of which VAT is deductible pursuant to Articles 168, 169 and 170, and for transactions in respect of which VAT is not deductible, only such proportion of the VAT as is attributable to the former transactions shall be deductible.

The deductible proportion shall be determined, in accordance with Articles 174 and 175, for all the transactions carried out by the taxable person….”

18. Articles 174 and 175 of the Directive stipulate that, subject to certain exceptions which are not relevant in this case, the deductible proportion is to be determined by a fraction of which the numerator is the turnover, exclusive of VAT, attributable to taxable transactions and the denominator is the turnover, exclusive of VAT, attributable to taxable transactions and the turnover attributable to exempt transactions.

19. The principles for calculating the deductible proportion of a taxable person’s input tax set out in Articles 174 and 175 of the Directive have been enacted into UK domestic law by Sections 24 to 26 of the VATA. So far as is material to this decision:

Section 24(1) of the VATA provides as follows:

“….“input tax”, in relation to a taxable person, means the following tax, that is to say—

VAT on the supply to him of any goods or services;…

being (in each case) goods or services used or to be used for the purpose of any business carried on or to be carried on by him.”

Section 25(2) of the VATA provides that a taxable person “is entitled at the end of each prescribed accounting period to credit for so much of his input tax as is allowable under section 26, and then to deduct that amount from any output tax that is due from him”;

Section 26(1) of the VATA provides that credit is to be given for “so much of the input tax for the period (that is input tax on supplies, acquisitions and importations in the period) as is allowable by or under regulations as being attributable to supplies within subsection (2) below”;

Section 26(2) of the VATA provides that:

“The supplies within this subsection are the following supplies made or to be made by the taxable person in the course or furtherance of his business—

taxable supplies;…”; and

Section 26(3) of the VATA provides that:

“(3) The Commissioners shall make regulations for securing a fair and reasonable attribution of input tax to supplies within subsection (2) above…”.

20. The regulations to which Section 26(3) of the VATA refers are the Regulations. At the times to which this decision relates:

paragraph 101(2) of the Regulations provided that, subject to various exceptions which are not material to this decision:

“…in respect of each prescribed accounting period—

(a) goods imported or acquired by and …goods or services supplied to, the taxable person in the period shall be identified,

(b) there shall be attributed to taxable supplies the whole of the input tax on such of those goods or services as are used or to be used by him exclusively in making taxable supplies,

(c) no part of the input tax on such of those goods or services as are used or to be used by him exclusively in making exempt supplies, or in carrying on any activity other than the making of taxable supplies, shall be attributed to taxable supplies,

(d) …subject to subparagraph (e) below, there shall be attributed to taxable supplies such proportion of the residual input tax as bears the same ratio to the total of such input tax as the value of taxable supplies made by him bears to the value of all supplies made by him in the period,

(e) the attribution required by subparagraph (d) above may be made on the basis of the extent to which the goods or services are used or to be used by him in making taxable supplies,…”;

(2) paragraphs 101(4) and 101(5) of the Regulations provide for the ratio calculated in accordance with sub-paragraph 101(2)(d) of the Regulations to be expressed as a percentage and, if that percentage is not a whole number, to be rounded up; and

(3) paragraph 101(10) of the Regulations provides that “[in] this regulation “residual input tax” means input tax incurred by a taxable person on goods or services which are used or to be used by him in making both taxable and exempt supplies”.

input tax recovery - the principle of attribution

Direct and immediate

21. It may be seen from the legislation set out above that input tax is deductible only to the extent that the cost to which the input tax relates is a “cost component” of the taxable person’s taxable supplies (see Article 1(2) of the Directive). Article 168 of the Directive expands on this phrase by identifying the input tax which is deductible as that which relates to goods or services “used for the purposes of” the taxable person’s taxable supplies. In BLP Group plc v Customs and Excise Commissioners (Case C-4/94) [1995] STC 424 (“BLP”), the Court of Justice of the European Communities (the “CJEU”) made it clear that the above language is satisfied as long as there is a “direct and immediate link” between the transaction to which the input tax relates and the taxable person’s taxable supplies – see BLP at paragraphs [19] to [21]. In particular, in The Commissioners for Her Majesty’s Revenue and Customs v Royal Opera House Covent Garden Foundation [2021] EWCA Civ 910 (“ROH”), the Court of Appeal made it clear that the reference to “cost components” in Article 1(2) of the Directive does not mean that the cost of the transaction to which the input tax relates needs to be reflected in the prices charged by the taxable person for its taxable supplies (see ROH at paragraph [17]). It said that both “cost components” and objectively determined “purposes” were very general terms which were encapsulated in the “direct and immediate link” test (see ROH at paragraph [18]). Accordingly, although, for the sake of consistency, we will refer throughout this decision solely to the test’s being one of a “direct and immediate link” between a cost and a supply, that phrase should be taken to be synonymous with the phrases “cost component” and “used for the purposes of”.

22. It is for the national courts to apply the “direct and immediate link” test to the facts of each case before them and to take account of all the circumstances surrounding the transactions in issue – see Midland Bank plc v Customs and Excise Commissioners (Case C-98/98) [2000] STC 501 (“Midland Bank”) at paragraph [25] and Dial-a-Phone Limited v Her Majesty’s Customs and Excise Commissioners [2004] EWCA Civ 603 (“DaP”) at paragraph [26].

23. In terms of domestic case law, the Court of Appeal has referred to this principle in a number of its decisions, including DaP at paragraph [14] and Mayflower Theatre v The Commissioners for Her Majesty’s Revenue and Customs [2006] EWCA Civ 116 (“Mayflower”) at paragraph [9].

24. The corollary of the formulation described in paragraphs 21 to 23 above is that:

no right to deduct input tax arises in circumstances where the input tax relates to costs which have a direct and immediate link only with the taxable person’s exempt supplies; and

where input tax relates to costs which have a direct and immediate link with both the taxable person’s taxable supplies and the taxable person’s exempt supplies, only the portion of the input tax which relates to costs attributable to the taxable person’s taxable supplies is recoverable.

Overheads

25. Certain costs incurred by a taxable person do not have a direct and immediate link with either the taxable person’s taxable supplies or the taxable person’s exempt supplies. This could be for various reasons.

26. One is that the transaction with which such costs have a direct and immediate link does not amount to a supply for VAT purposes at all – see Abbey National plc v Her Majesty’s Customs and Excise Commissioners (Case C-408/98) [2001] STC 297 (“Abbey National”) (where the costs in question related to a transfer as a going concern), Kretztechnik AG v Finanzamt Linz (Case C-465/03) [2005] STC 1118 (where the costs in question related to a share issue) and Cibo Participations SA v Directeur regional des impôts du Nord-Pas-de-Calais (C-16/00) [2002 STC 460 (“Cibo”) (where the costs in question related to the acquisition of shares in a subsidiary). In each of these cases, the relevant costs related to a transaction which did not involve the making of a supply for VAT purposes.

27. Alternatively, the absence of a direct and immediate link to either category of supplies may be attributable to the fact that the costs are too general in nature to be capable of being linked directly to any specific supply or any specific supplies. Examples of this given by Lord Millett in Her Majesty’s Customs and Excise Commissioners v Redrow Group plc [1999] STC 161 (“Redrow”) were “[audit] and legal fees and the cost of the office carpet” (see Redrow at 169h).

28. Costs which do not have a direct and immediate link with either the taxable person’s taxable supplies or the taxable person’s exempt supplies are commonly referred to in the VAT context as “overhead costs”. The input tax relating to overhead costs is deductible where the costs are attributable to the taxable person’s economic activity as a whole (but only in the proportion that the taxable supplies made in the course of that economic activity bears to the aggregate supplies made in the course of that economic activity) – see Mayflower at paragraphs [26] to [34]. The basis for the deduction in that case is that the costs in question have a direct and immediate link with the taxable person’s business as a whole and, as such, are “cost components of an undertaking’s products” (see Cibo at paragraph [33]).

relevant case law

29. As we have observed in paragraph 22 above, the CJEU held in Midland Bank that it is for the national court to determine when a cost has a direct and immediate link with a particular supply and that, in doing so, the national court must take into account all the circumstances of the particular case. This means that, in the case of each taxable person, the question of identifying the supply or supplies to which a particular cost has a direct and immediate link is a mixed question of fact and law – see DaP at paragraph [31]. It also means that the authorities in this area do no more than establish certain broad principles of law which need to be followed in each case and that each case needs to be decided on its own particular facts.

30. In the UK, the question of attribution has been addressed in four seminal Court of Appeal decisions – Her Majesty’s Customs and Excise Commissioners v Southern Primary Housing Association Limited [2003] EWCA Civ 1662 (“Southern”), DaP, Mayflower and, most recently, ROH.

31. The earliest of these decisions is Southern. The facts in Southern were, in summary, as follows. The appellant, a developer, purchased a piece of land on which it incurred input tax. It sold the land on to a housing association and entered into an agreement with the housing association under which it was to develop the land by building houses on it. It was common ground that the transactions were commercially connected and that the purchase of the land was a supply for VAT purposes. The appellant contended that it was entitled to deduct the input tax which it had paid on its purchase of the land from its output tax, on the basis that there was a direct and immediate link between the cost of the land and the taxable supplies which it made in developing the land. The Respondents rejected that contention and, although the appellant succeeded in its appeal against that conclusion in both the VAT Tribunal and the High Court, the Respondents prevailed before the Court of Appeal.

32. The Court of Appeal accepted that the land purchase was commercially necessary in order for the appellant to make its supplies of developing the land. However, the land purchase did not have a direct and immediate link with those supplies, unlike the cost of the materials which the appellant needed to purchase in order to carry out that development. There was a link between the cost of purchasing the land and the supplies under the development contract, but that link was not direct or immediate. The development contract would not have been made but for the associated land purchase. But “but for” was not the test and did not equate to the “direct and immediate link” test.

33. Jacob LJ noted that the tribunal had erred by adopting too generalised an approach to the transactions which had occurred. He said that “[you] have to look at transactions individually, component transaction by component transaction. They may be linked in the sense that one would not have happened without the other, but they remain distinct transactions nonetheless. Only if one transaction is merely ancillary to a main transaction can one disregard the distinct nature of each transaction . . .. If that were not so, the principle of neutrality would be violated. Moreover, there would be intractable problems as to which input was being attributed to which part of the 'overall transaction'. You may find, as here, taxable and exempt transactions all mixed up in the same 'overall' transaction – which is illegitimate.”

34. Thus, Southern is authority for the proposition that the mere fact that “but for” the cost in question, the particular supplies would not have been made is not enough to establish the requisite direct and immediate link between the cost in question and the relevant supplies.

35. The appellant in DaP sought to rely on the decision in Southern to support its submission that there was no direct and immediate link between certain advertising costs which it had incurred and its exempt supplies of insurance intermediary services. In DaP, the appellant made taxable supplies of introducing customers to airtime service providers and mobile phone networks through airtime service contracts and exempt supplies of insurance intermediary services. Its adverts for the airtime service contracts referred to the fact that customers could take advantage of an initial three months of free insurance but the adverts did not refer to the opportunity to continue with the insurance thereafter. The appellant did not become entitled to its commission from the insurer until the customer continued to be insured after the three-month free period. The Court of Appeal rejected the appellant’s approach. It approved of the conclusion drawn by the VAT Tribunal in that case to the effect that the cost of the adverts had a direct and immediate link with both the appellant’s taxable supplies and the appellant’s exempt supplies. It agreed with the tribunal that the free insurance had been used not merely to attract customers to sign airtime service contracts but was clearly intended to attract customers to the insurer and that the appellant had a direct financial interest in customers’ staying with the insurer on completion of the free three months. Jonathan Parker LJ described this conclusion as amounting to “no more than the drawing of well-nigh irresistible inferences from the undisputed facts” (see DaP at paragraph [73]).

36. Jonathan Parker LJ said that, in this regard, it did not matter that the supplies of insurance intermediary services might be viewed as being in a commercial sense secondary to the making of the supplies of airtime service contracts or even that they might be made only after the supplies of airtime service contracts were made. All that mattered was whether there was a direct and immediate link between the relevant costs and the supplies in question.

37. In any event, he went on to reject the proposition that the supplies of insurance intermediary services were secondary to the supplies under the airtime service contracts. He pointed out that the appellant’s entitlement to commission from the insurer “represented a substantial proportion of its income during the relevant period. Insurance with Cornhill was part of the package of services advertised by DaP. To my mind, the fact that the advertisements referred only to the initial free three-month period of insurance is hardly surprising, and says nothing as to the relative importance to DaP of the insurance element of the package as compared with the making of taxable supplies” (see DaP at paragraph [77]).

38. He also rejected the proposition that the insurance intermediary services were provided only after the taxable supply to the relevant customer had been made, saying that the correct way to view the situation was that the insurance intermediary services were supplied by the appellant to the insurer on a continuous basis and therefore the precise order in which the taxable and exempt services were supplied by the appellant vis-à-vis any particular individual customer was entirely beside the point.

39. The decision in DaP is of particular relevance to these appeals because it also related to advertising costs and it was a case where the Court of Appeal found that there was a direct and immediate link between those advertising costs and two different categories of supply made by the taxable person who had incurred the costs, which is the primary argument made by the Respondents in this case.

40. The third decision of the Court of Appeal which is relevant to these appeals is the decision in Mayflower. That case related to the costs incurred by the appellant in buying in theatrical productions. The appellant claimed to recover part of the input tax on those costs on three alternative bases. One of those bases was that the relevant costs had a direct and immediate link with both the appellant’s exempt supplies and the appellant’s taxable supplies and another was that the relevant costs had a direct and immediate link with the business as a whole so that they were overheads. (The third of the alternative bases is of no relevance to these appeals).

41. The Court of Appeal rejected the second argument. After describing the reasoning which had featured in the earlier overheads cases referred to in paragraph 26 above, Carnwath LJ rejected the proposition that the nature of a direct and immediate link between a cost and an ensuing supply could vary from business to business and encompassed a spectrum of possibilities from the direct attribution of a given input to a given output, at one end, to overheads of the business attributable to the whole business activity at the other. He said, at paragraph [33]:

“In my view, the metaphor of the ‘spectrum’ is unhelpful; a ‘slippery slope’ might be more apt. The special treatment of ‘overheads’ or ‘general costs’ serves a particular and limited purpose in the VAT system, for those inputs which would not otherwise be brought within the calculation. It should not be extended beyond that purpose.”

42. As regards the first argument, the Court of Appeal upheld the decision at first instance to the effect that the links between the costs of buying in the productions and both the services provided by the appellant in selling the production companies’ merchandise and the appellant’s supplies of sponsorship were not sufficiently direct and immediate for the appellant to succeed by reference to those supplies. However, it accepted that the costs had a direct and immediate link to the taxable supplies of programmes made by the appellant. This was because the productions were part of the content and subject matter of the programmes and therefore the cost of buying in the productions had a direct and immediate link with the programmes in precisely the same way as the cost of the ink and paper which were used to create the programmes.

43. Mayflower is significant in that it shows that costs should not be allocated lightly to the category of overheads. That category is residual and a cost should be allocated to it only in circumstances where it is impossible to identify a direct and immediate link between the cost and any supply or supplies made by the taxable person.

44. The most recent of the relevant Court of Appeal decisions is ROH. As had been the case in Mayflower, ROH related to production costs although, in ROH, the productions were put on by the appellant itself as opposed to being bought in. The matter which fell to be decided in ROH was whether those production costs had a direct and immediate link solely to the exempt supplies of tickets which were made by the appellant (as the Respondents contended) or had a direct and immediate link to both the exempt supplies of tickets and the taxable supplies of catering which were made by the appellant at the time of the performances (as the appellant contended). The facts found at first instance in the case were that there was a close economic link between the production costs and the catering supplies in that:

the production costs were essential to enable the appellant to make the catering supplies (because they enabled the appellant to put on the productions which attracted the customers to its catering supplies);

the catering supplies assisted in giving the customers a “fully integrated visitor experience”; and

the revenue from the catering supplies was used to defray the production costs.

45. Despite those close economic links, the Court of Appeal upheld the Upper Tribunal’s decision in favour of the Respondents. It held that the relevant test was not whether there was a close economic link between the production costs and the catering supplies but rather whether the link between the production costs and the supplies was direct and immediate. It said that, even though the production costs had been found to be essential to the making of the catering supplies, that was no more than an expression, in another form, of the “but for” test which had been held in Southern to be a necessary but insufficient condition for the establishment of a direct and immediate link. The link between the production costs and the catering supplies in ROH was only indirect because the production costs were physically used solely for the purpose of putting on the performances. As such, they had a direct and immediate link solely to the ticket sales and, following the earlier decision in Mayflower, to the programmes in which material from the performances was directly reflected. However, the costs were not physically used in making the catering supplies. The fact that the catering supplies would not have been made but for the performances and, hence, but for the production costs, was not enough to create a direct and immediate link between the production costs and the supplies. The link was only indirect.

46. ROH is of significance in this case not only because it is the most recent occasion on which the Court of Appeal has addressed the direct and immediate link test but also because it demonstrates that a close economic link will not always amount to a direct and immediate link.

47. Turning to some relevant first instance decisions:

in Britannia Building Society v The Commissioners of Her Majesty’s Customs and Excise [1997] BVC 4106 (“Britannia”), television adverts referring to specific financial products were held to have a direct and immediate link with both taxable and exempt supplies because the VAT Tribunal found that none of them was designed to illustrate any particular aspect of the appellant’s services but they were rather produced with the intention of promoting the appellant’s brand more generally;

in Royal Agricultural College v The Commissioners of Her Majesty’s Customs and Excise [2002] Lexis Citation 535 (“RAC”), the VAT Tribunal held that the cost of marketing materials did not have a direct and immediate link with supplies of conference facilities and catering because the purpose of the marketing was to recruit students to the college. The link with any subsequent supply of conference facilities and catering to the students was indirect and, in any event, even if that had not been the case, the intervening exempt supplies of education were sufficient to break the link between the cost of the marketing materials and the supplies of conference facilities and catering. The VAT Tribunal said (at paragraph [42]) that:

“The direct and immediate link is clearly that of attracting students to the College. The link that thereby they contribute to the College's taxable activities such as, for example, using the bar, is indirect and not immediate, in the sense in which that term is used. If it has no students it will not be successful in its wider activities”;

in Skipton Building Society v The Commissioners for Her Majesty’s Revenue and Customs [2009] UKFTT 191 (TC) (“Skipton”) the First-tier Tribunal held that costs of print advertising had a direct and immediate link with exempt supplies of mortgage services in those cases where the mortgage services were mentioned overtly in the advert (even if only in passing) but not in those cases where that was not the case;

in DFS1, to which we referred briefly in paragraph 6 above, the First-tier Tribunal held that traditional advertising costs did not have a direct and immediate link with supplies of insurance intermediary services. In reaching its conclusion, the First-tier Tribunal held that the content of the adverts was relevant but not determinative. It said that “use” was not a physical use but some form of economic use. For example, an advert might mention sprats when its purpose was to attract customers to purchase mackerels. In each case, it was necessary to look at all of the facts and circumstances objectively;

in N Brown Group PLC; JD Williams & Company Ltd v The Commissioners for Her Majesty’s Revenue and Customs [2019] UKFTT 172 (TC) (“N Brown”), the First-tier Tribunal held that advertising costs had a direct and immediate link with supplies of credit provided on the sale of clothes even though the availability of that credit was not mentioned in the advertising itself. In reaching that conclusion, the First-tier Tribunal held that each of:

the physical content of the adverts;

the subjective intention of the individuals acting for the taxable person in choosing the content of the advertising;

the way that prospective customers might respond to that content; and

the way that the taxable person chose to measure the success of its marketing or set its marketing budget

were all relevant but not determinative. In each case, the focus was on objectively determining the use of the advertising costs by the taxable person, taking into account all the facts and circumstances. In N Brown, there was found to be a two-way relationship between the supplies of goods and the supplies of credit in that an increase in supplies of goods led inexorably to an increase in supplies of credit and, at the same time, the availability of the credit stimulated an increase in supplies of goods. The fact that there was such a predictable correlation between the supplies of goods and the supplies of credit meant that there was no need for the appellant to monitor the effect of its adverts on supplies of credit specifically and therefore the fact that it did not do so carried little weight.

In addition, the First-tier Tribunal rejected the parallel which the appellant sought to draw in that case between the cost of purchasing the raw material for the goods (for example, the purchase of a black dress) and advertising costs. In rejecting the submission by the appellant in that case that, if there was a direct and immediate link between the advertising costs and the supplies of credit, the same direct and immediate link must be found to exist between the cost of the black dress and the supplies of credit, the First-tier Tribunal said that there was a difference between the application of the test in the context of advertising and the application of the test in the context of other costs of the business. It said:

“[We] think the answer is to be found in considerations of “use”. Marketing material is “used” to stimulate demand for economic facilities that the Group is offering and the task is to identify whether it is stimulating demand only for retail goods (as the Group argues) or for both goods and credit (as HMRC argue). In order to understand what the Group’s marketing material is promoting, it is necessary to engage in a detailed analysis of the Group’s business and how its credit offering interacts with its goods offering. However, the Group does not use the costs of purchasing a black dress to promote anything. Rather, it uses those costs as part of a straightforward transaction of purchase and sale and no wider examination of the Group’s business is necessary to explain the conclusion that the costs of purchasing the black dress are used, and used only, to enable the black dress to be sold at a profit”; and

in The Roald Dahl Museum and Story Centre v The Commissioners for Her Majesty’s Revenue and Customs [2014] UKFTT 308 (TC) (“Roald Dahl”), the appellant submitted that its exhibition costs had a direct and immediate link with both the exempt supplies of entry to the museum and the taxable supplies of sales in the museum shop. The Respondents submitted that the costs had a direct and immediate link solely with the exempt supplies. The First-tier Tribunal held that there was no direct and immediate link between the costs and those items sold by the museum shop which were also sold by shops unrelated to the museum and could be purchased by members of the public in the museum shop without accessing the museum exhibits. The mere fact that it would be rare for someone to purchase such items without also accessing the museum exhibits or that the sales of such items would have been lower if the museum exhibits had not existed did not mean that there was a direct and immediate link between the costs and the sales of such items. Similarly, the exhibition costs did not have a direct and immediate link with supplies of:

postcards which used the same ideas as were used for displays in the museum but did not have those displays as their subject matter; and

postcards which had on them photographs of items displayed in the museum.

In both cases, the link between the costs and the relevant supplies was indirect in nature.

However, the same was not true of a so-called “Hut Book” (a guide which had been produced both for the purpose of interpreting objects in the museum and for sale in the shop).

48. The case law cited to us in argument also included certain cases dealing with the apportionment of input tax in circumstances where it was already accepted that the cost to which the relevant input tax related either had a direct and immediate link with both taxable and exempt supplies made by the relevant appellant or amounted to an overhead of the relevant appellant’s partially-exempt business.

49. One of those cases was St Helen’s School Northwood Ltd v The Commissioners for Her Majesty’s Revenue and Customs [2007] STC 633 (“St Helen’s”). The issue between the parties in that case was whether the basis of apportionment between taxable and exempt supplies proposed by the appellant in relation to the costs of building a sports complex would give rise to a fair and reasonable rate of input tax recovery. In dismissing the appellant’s appeal, Warren J held that the method of apportionment which the Respondents were proposing instead better reflected the economic use which was being made of the sports complex to which the relevant expenditure related than did the method proposed by the appellant. The taxable use of the sports complex was essentially a secondary use and the main use was the exempt use. In reaching that conclusion, Warren J held that:

the fact that prior case law had held that motive and purpose were not relevant to the apportionment process did not allow a tribunal to disregard the observable terms and features of the transaction and the wider context in which the transaction came to be carried out;

physical use might reflect economic use but did not necessarily do so; and

any apportionment must give a credible result in economic terms

(see St Helen’s at paragraphs [60], [63] and [75] to [77]).

50. In The Commissioners for Her Majesty’s Revenue and Customs v London Clubs Management Limited [2011] EWCA Civ 1323 (“London Clubs”), the appellant was the representative member of a group which operated casinos. It acquired premises with greater floor space which it used to make catering and gaming supplies and its input tax was initially apportioned on the basis of turnover. It proposed a revised basis of apportionment based broadly on the use of the floor space and its appeal against the Respondents’ refusal of that proposal was successful before each court up to and including the Court of Appeal. Etherton LJ, with whom both Pitchford and Ward LJJ agreed, noted that the process of apportionment was relevant only where an item was a cost component of two different supplies, one exempt and one taxable. He went on to approve of the approach to apportionment which had been adopted by Warren J in St Helen’s (see paragraph [38]).

51. In The Commissioners for Her Majesty’s Revenue and Customs v Lok’nStore Group plc [2014] UKUT 288 (TCC) (“Lok’nStore”), the appellant made taxable supplies of, inter alia, storage space and exempt supplies of insurance. A question arose as to how the input tax on its overheads should be apportioned between the two categories of supplies. The appellant proposed a special method based on a mix of floor space and values whilst the Respondents maintained that the standard method of apportionment based on turnover should apply. The Upper Tribunal held that the appellant’s special method better reflected the economic use of the overheads than did the Respondents’ standard method. In doing so, the Upper Tribunal was critical of the approach which had been adopted by the First-tier Tribunal in that case, which it said failed adequately to distinguish between the two separate stages in the process of determining input tax recovery – that is to say, first, the question of attribution and, secondly, the question of apportionment once attribution had occurred. In the words of the Upper Tribunal, the question of attribution was a “logically prior question” to the question of apportionment (see paragraph [21]) and it should play no part in the apportionment process. However, depending on the facts, considerations which were relevant at the attribution stage might also be relevant at the apportionment stage.

ppc advertising

52. We now turn to setting out the evidence and facts in relation to the appeals.

53. However, before doing so, we think that it would be helpful to say something about the two distinct types of PPC advertising with which these appeals are concerned. Those are:

“shopping” adverts, in which the input of a search term by a user leads to photographs on the Google search page of specific products from the relevant retailer’s website; and

“search” adverts, in which the input of a search term by a user leads to text adverts in relation to the relevant retailer on the Google search page.

In both cases, by clicking on the relevant photograph or text advert, the user is taken to a page on the retailer’s website which is called a “landing page”. From the landing page, the user is then able to navigate to other pages on the retailer’s website, as desired.

54. Before using Google’s PPC advertising services, a retailer sets up an account with Google which, in some case, contains specific objectives and goals. Once a goal has been chosen, Google then provides features and settings to help the retailer obtain the results it wants. Goals can be changed periodically to reflect changes to the retailer’s business objectives.

55. For Google “search” adverts, the retailer identifies the keywords which it wishes to use and participates in an auction through Google in which it bids against other retailers for the use of the same keywords. The more that a retailer is willing to pay for each click on its advert, the more likely its advert is to appear in the search results (although Google also takes into account other factors such as the relevance and usefulness of the advert to the prospective user). Google does not get paid unless a user clicks on the advert. The mere fact that the advert appears on Google as an impression does not generate a fee.

56. There are broadly three different types of “search” advert. These are as follows:

“generic”, which involves the use of a key term relating to a specific product – for example, in this case, a term such as “sofa” or “corner sofa”;

“brand”, which involves the use of the retailer’s own name – for example, in the case of Sofology, a term such as “Sofology” or “Sofology two-seater” and, in the case of DFS, a term such as “DFS” or “DFS two-seater”; and

“competitor”, which involves the use of the names of the retailer’s competitors – for example, in the case of Sofology, a term such as “DFS” or Ikea” and, in the case of DFS, a term such as “Sofology” or “Ikea”.

57. Of these three types, the generic “search” adverts tend to be the most expensive as there are more retailers bidding for them in the auction. The broader the term, the more expensive the term is. There is less demand for the brand and competitor “search” adverts and they are therefore less expensive than the generic “search” adverts.

the evidence

Introduction

58. The evidence in relation to the appeals took the form of:

various documents and screenshots;

a live demonstration on screen at the hearing of the impact of clicking on certain adverts;

some videos which had been included on each Appellant’s website; and

written and oral witness evidence.

59. The latter involved the evidence of ten witnesses on behalf of the Appellants –

Mr James Robinson – the head of conversion rate optimisation at Sofology;

Mr Peter Kennedy – a store manager at Sofology;

Ms Janet Duckworth – a former marketing manager at Sofology and now head of marketing at Sofology;

Mr Sebastian Brown – formerly a senior marketing and creative manager at Sofology and now director of creative at Sofology;

Mr Peter McDonald – the former head of finance at Sofology and now head of investor relations and group financial planning and analysis for the group including both Sofology and DFS;

Mr James Brewer – the digital and brand marketing director of DFS;

Mr James Vernon – the head of online at DFS;

Ms Clare Johnson – a regional manager at DFS;

Mr Nick Ashworth – the former head of marketing and marketing director of DFS; and

Mr Nick Smith – the chief executive officer of DFS.

60. Each of the witnesses executed a written witness statement and each of them, apart from Mr Brown, was cross-examined by the Respondents’ counsel at the hearing.

61. We set out in the paragraphs which follow our summary of the evidence with which we have been provided.

62. That summary can most usefully be set out under the following seven broad headings in relation to each Appellant:

the contractual arrangements between the relevant Appellant and its customers, the contractual arrangements between the relevant Appellant and the insurer to whom it was providing insurance intermediary services and the contractual arrangements in relation to the relevant Appellant’s PPC advertising on Google;

the objectives of the PPC advertising taken out by the relevant Appellant during the relevant period;

the nature of that advertising;

the journey taken through the website by a user upon clicking on a Google “shopping” or “search” advert placed by or on behalf of the relevant Appellant;

the training delivered by the relevant Appellant to its sales staff during the relevant period so far as it pertained to sales of sofa insurance;

the manner in which sofa insurance was sold online, in-store and on the telephone to the customers of the relevant Appellant during the relevant period; and

certain financial information in relation to the relevant Appellant during the relevant period, showing, inter alia, the importance of revenue deriving from the relevant Appellant’s services as an insurance intermediary relative to the relevant Appellant’s other income.

Sofology

Contractual arrangements

Contractual arrangements with customers

63. During the Sofology relevant period, Sofashield was not a product which was sold in isolation and a customer could not purchase Sofashield without purchasing a sofa. A customer could purchase Sofashield only:

at the same time as he or she purchased the relevant sofa; or

at any time up to delivery of the sofa.

However, Sofashield was available on virtually all of the sofas sold by Sofology during the Sofology relevant period. The only exclusions were sofas previously returned by customers and sofas used in commercial premises. At the hearing, Mr McDonald accepted that those were small in number.

Contractual arrangements with Castelan

64. We were not provided with the contract between Sofology and its insurer which was in force for most of the Sofology relevant period. However, we were provided with a contract between the two companies dated 23 September 2016, which covered the final month of the Sofology relevant period and, in a board pack prepared at the time when that contract was executed, Castelan was referred to as “an incumbent supplier”. Accordingly, we have proceeded on the basis (which we consider to be reasonable) that the contract in relation to the insurance intermediary services for the earlier part of the Sofology relevant period was also with Castelan and broadly followed the same form.

65. Under the terms of the agreement:

Sofology was obliged to promote, market and sell Sofashield to its customers in connection with its sales of sofas;

the more policies Sofology was able to sell to its customers, the lower the cost per policy was to Sofology; and

there was a significant difference between the amount charged by Sofology to the customer and the amount payable to Castelan by Sofology with the result that, even after taking into account the fact that the former amount included insurance premium tax, Sofology made a significant profit out of each policy which it sold.

Contractual arrangements in relation to PPC advertising on Google

66. Under the terms of the contract between Sofology and Google dated 10 September 2013:

Sofology authorised Google to place Sofology’s advertising materials and related technology on Google;

Sofology was solely responsible for, inter alia, the creative aspects of the adverts, ad-trafficking and targeting decisions, keywords and landing pages; and

charges were based on the applicable billing metrics (in this case, clicks and not impressions).

67. Thus, under the terms of the contract between Sofology and Google, Google was obliged to place Sofology’s advertising materials on Google, Sofology was not obliged to pay Google unless and until a person clicked on its advert and Sofology had control over the identity of the landing page to which the person clicking on the advert was directed.

Objectives

68. In his witness statement, Mr Brown said that:

none of the marketing strategy or spend was directed specifically for the purpose of increasing the sales of Sofashield;

Sofashield did not drive any aspect of the determination of Sofology’s marketing strategy “because the marketing strategy is solely centred around the sofas and ensuring customers find the right sofa that suits their needs”; and

Sofashield did not have any bearing on a customer’s decision-making process in terms of purchasing a sofa and therefore, “from a marketing perspective, we do not treat Sofashield as ever being a relevant consideration by a customer in making a determination as to whether to buy a sofa product”.

69. In keeping with that approach, Ms Duckworth said that the company “[used] Google ads strategically to ensure that where customers are searching for a sofa product…we appear at the top of a customer’s search results on Google for consideration…The sofa products are the key driver in deciding on the ‘ad words’ that Sofology bids for with Google…”. Ms Duckworth added that “[no] part of the marketing spend is allocated to Sofashield to try to increase sales of the Sofashield offering”. In her view, Sofashield was simply an offering to customers who had already decided to purchase a sofa and it was “not a consideration for us in setting the marketing strategy around the customer journey”. Ms Duckworth explained that the marketing strapline used by Sofology over the relevant period was the phrase “Feel at home on the sofa you love” and that the relevant phrase was being used as a touchstone or guiding principle to inform the marketing strategy.

70. Although the marketing strategy was not specifically directed at selling Sofashield, the evidence suggests that the position was a little more nuanced than either Mr Brown or Ms Duckworth stated.

71. First, Ms Duckworth accepted that the phrase “Feel at home on the sofa you love” was used by the company not just in relation to sales of sofas but also in relation to sales of Sofashield. In Ms Duckworth’s view, Sofashield was a service which, like the sofas themselves, enhanced the Sofology brand. She said that a customer would feel more at home on the sofa he or she had purchased because it was known that the sofa would be kept clean and mended and would last longer. Consistent with that approach, we were shown:

a Sofashield leaflet which was handed out to in-store customers following their purchase and which used the strapline; and

internal company training materials in which the strapline was linked to Sofashield in the following manner:

“‘Feeling at home on a sofa they love’. How can they do this without Sofashield?”; and

“Sofashield Tip 17 – “Living with small marks and scratches is not “feeling at home on a sofa you love”. Sofology believes in Sofashield. We believe that every customer needs it and that it will make them love their sofa more.”

72. Secondly, the documentary evidence suggested that the marketing strategy was not specifically directed solely at sales of sofas per se. At the hearing, we were provided with two presentations to the company from the media agency which has managed and run the company’s PPC advertising campaigns since 2017. Both of those presentations referred to the fact that that the primary objective of the company’s PPC advertising strategy was to increase footfall to stores. The key performance indicators (or “KPIs”) which were used to monitor the attainment of that objective, as set out in the presentations, were not expressed to be limited specifically to sales of sofas, as distinct from sales of sofa insurance, but were instead general in nature. The KPIs referred to in one of the presentations were “Web Sales, Google store visits, onsite store locator usage, Sofology instore sales data”.

73. Thirdly, although both Ms Duckworth and Mr Robinson, in their evidence, initially said that the ultimate aim of the marketing strategy was to drive footfall into stores in order to achieve sofa sales, Mr Robinson accepted that the reason why it was important to drive footfall into stores was that this increased sales of both sofas and sofa insurance (because the number of customers buying eligible sofas who also took up sofa insurance – which the parties referred to at the hearing as the “attachment rate” – was much higher for sales in-store than it was for sales on the telephone or online).

Nature

74. The balance of the nature of Sofology’s PPC advertising changes from time to time, as one would expect with an ongoing dynamic marketing strategy.

75. Mr Robinson testified that:

given the greater expense involved in bidding for generic “search” adverts, the company tended to prioritise brand or competitor “search” adverts over generic “search” adverts in allocating its marketing budget although the precise breakdown of the marketing spend on Google adverts as between “shopping” adverts and each of the three different categories of “search” adverts varied from time to time;

the company had not been bidding on generic “search” adverts more recently; and

brand “search” adverts were not designed to raise awareness of the Sofology brand as, by definition, the user would already be aware of the brand in using the search term. Instead, the reason why brand “search” adverts were so important was because they were used to defend Sofology's territory online by seeking to ensure that customers clicked on the Sofology link and went through to its website rather than the website of a competitor and also visited its stores.

76. The company did not use “shopping” adverts until the final month of the Sofology relevant period, given that it was not possible to make purchases online until that time. As such, all but a small fraction of the PPC advertising during the Sofology relevant period took the form of “search” adverts. (For example, Google analytics data provided by the company showed that, over the last twelve months of the Sofology relevant period, there were 44,430 clicks on “shopping” adverts as compared to 1.56 million clicks on brand “search” adverts).

77. We were not provided with a breakdown for the Sofology relevant period of the overall spending on “search” adverts as between the three different categories of “search” adverts although Mr Robinson provided a breakdown between the three categories for the last six months of 2019. This showed that, of the aggregate spending on PPC advertising within that period which was allocable to “search” adverts (56% of the total spend), approximately 48% was spent on generic “search” adverts, approximately 29% was spent on brand “search” adverts and approximately 22% was spent on competitor “search” adverts.

78. Mr Robinson said that the company did not bid on the term Sofashield at all “because the selection of search terms is based on terms that have the potential to lead to a conversion from the search to the sale of a sofa and the business does not treat Sofashield as a driver for making sales of sofas online. Further, Sofashield is not actively marketed as an offering by Sofology because it is not considered part of the offering by the business but as an added extra only once a sofa has been chosen for purchase.”

The journey through the website

79. Each of the “shopping” adverts contained photographs of sofas and each of the “search” adverts referred to sofas as part of the text. Neither category of advert referred to the availability of Sofashield.

80. As for the journey which occurred when a user clicked on an advert, Mr Robinson testified that:

the significant majority of the brand “search” adverts and the competitor “search” adverts led the user to the Sofology home page as the landing page; and

the majority of the generic “search” adverts led the user to either the Sofology home page or a broad category page (as opposed to an individual product page) as the landing page.

81. Both the home page and each broad category page (as well as each individual product page) showed images of sofas and made no reference to Sofashield. However, there was a navigational toolbar at the top of the relevant pages and one of the items on that toolbar was entitled “More”. By holding the cursor over the word “More”, the user was able to see a menu of miscellaneous items, one of which was Sofashield. The user could then click on Sofashield to get to the Sofashield page on the website. That page contained information about Sofashield and may also have contained a link to a promotional video. (There was a link to a promotional video on the page in August 2017 but Mr Robinson was not sure whether that link was also in place during the Sofology relevant period).

82. Where the landing page was the home page, it generally required two clicks to get from the home page to an individual product page – one click to get to a broad category page and then another to get to the individual product page – although, in some cases, an individual product featured in the navigational toolbar so that its page could be reached in one click from the home page. In contrast, the Sofashield page could be accessed with one click from the home page.

83. In addition to the Sofashield page mentioned above:

an option to add Sofashield to the customer’s basket also appeared on the website once the relevant sofa had been added to the basket (apart from in the rare cases mentioned in paragraph 63 above) although, as we have previously noted, no purchases could be made online until the final month of the Sofology relevant period; and

Mr Robinson explained that there were also two articles on Sofashield in the “Help” section of the website.

84. Each individual product page contained details of other items of furniture in which the visitor might be interested, to which the visitor was able to navigate by clicking on the relevant item.

Training

85. The in-house training delivered to Sofology sales staff placed a significant emphasis on sales of Sofashield. Sales staff were trained to obtain (or “bank”) key information on the relevant customer’s lifestyle from the start of interacting with the customer so that the lifestyle information could be used to sell Sofashield at the point when the customer had decided on his or her sofa purchase. For example, we were taken to a “Top tips” document for sales staff which noted that:

“Customers will easily agree that they eat and drink on their sofas early in the sale when you relate it to choosing the right leather or fabric and durability. Everyone does it. Get agreement early and bank it for later...Once a sofa is chosen and the ordering agreed, it’s easy to recall the eating and drinking and to make the service essential. Your customers will love that you were listening and accept that it’s needed more easily.”

86. In addition, the training guidance dated January 2016 instructed staff automatically to include Sofashield on the customer’s invoice and then, after informing the customer that that was the case, to remove Sofashield only if the customer insisted that that be done. (We were not provided with training materials for any period preceding January 2016. We were, however, provided with training materials for the period from December 2016 (after the end of the Sofology relevant period) which showed that that “opt-out” practice had ceased by that date).

87. A telephone script dated October 2014 showed that the same “opt-out” practice applied in the case of telephone sales. We were not provided with another telephone script after October 2014 until March 2018, which was after the end of the Sofology relevant period.

88. Moreover, sales staff were taught how best to counter any objections which a customer might make to the inclusion of Sofashield (or, as the training materials put it, how “to confidently overwhelm objections”) at that stage. Those training materials explained that:

“Your job begins when a customer says “no”. No doesn't mean never...it means not yet. "Yes" is the destination and "no" is how to get there. Don't give in [too] early! Get at least 3 [nos] before you give up! ([This] rule is subject to your rapport with your customer and the better your relationship will allow you to handle the objection for a longer of period of time without [affecting] the [customer’s] experience in store).”

89. The “Top tips” document informed staff not to mention Sofashield until the customer had decided to make the purchase. This was to reduce the risk of losing the sofa sale because, if the customer were to be told about the insurance early on, the customer would then take into account the cost of the insurance in considering whether or not to purchase a sofa. Thus, the key message in each case was “Sell the furniture first”. Having said that, Mr Kennedy said that, although sales staff were encouraged not to discuss Sofashield too early in the negotiations, they were always prepared to discuss Sofashield with a customer “if the conversation goes that way”.

Selling methods

90. We have already noted in paragraph 2 above that there were no online sales by Sofology until the final month of the Sofology relevant period. As regards such online sales as there were in that period, the customer was given the option to add Sofashield to the virtual basket only after placing the sofa in the virtual basket in the course of the checkout process. At that stage, the customer was able to add other additional items to the order in addition to Sofashield.

91. A significant proportion of sales during the Sofology relevant period took place in-store. In relation to such sales, Mr Kennedy explained that, once the customer had chosen a sofa, but before the sale was completed, the sales-person would offer the customer various add-on options, one of which was Sofashield.

92. For in-store sales, the customer was also given a leaflet relating to Sofashield after completing his or her purchase, regardless of whether or not the customer had chosen to take Sofashield. Ms Duckworth explained that it was open to a customer to purchase Sofashield at any time until the sofa was delivered.

93. Mr Kennedy said that sales staff were not incentivised to sell add-ons, including Sofashield.

94. As regards sales by telephone, the script used by the Sofology operators from October 2014 revealed that operators took steps to sell Sofashield, on an “opt-out” basis, along with professional care kits, after agreeing the sofa sale.

Financial information

95. As it was not possible for a Sofology customer to make a purchase online until the last month of the Sofology relevant period, the great majority of sales in the Sofology relevant period (more than 90% of sales) were made in-store and virtually all of the rest were made on the telephone.

96. The attachment rate in-store during the Sofology relevant period was 53.89%. This was in keeping with the general Sofology attachment rate for in-store sales between January 2016 and August 2020, which Mr McDonald testified was between 50% and 61%. (Attachment rates are always higher for in-store sales than telephone or online sales. Over the Sofology relevant period, the attachment rate for telephone sales was 44.02% and, following the Sofology relevant period, in 2019, the respective attachment rates were 56.07% for in-store sales, 36.14% for telephone sales and 14.13% for online sales.)

97. The revenue derived by Sofology from its provision of insurance intermediary services was taken into account in the same way as its other revenue when it came to determining the company’s marketing budget. As Ms Duckworth explained, this meant that:

a reduction in sales of Sofashield would lead to a reduction in the marketing budget; and

since the marketing budget was a significant factor in driving sales, a reduction in the marketing budget would, assuming all other market factors remained constant, result in a reduction in the sales of all of Sofology’s products, including sofas.

98. Sofashield made a significant contribution to the company’s gross sales and gross profit over the Sofology relevant period. Sofashield accounted for 5.54% of the company’s gross sales and 9.42% of the company’s gross profit over that period. The fact that the percentage of gross profit was so much higher than the percentage of gross sales reflected the greater profit margin implicit in sales of Sofashield. Mr McDonald accepted that Sofashield was “a high percentage margin product”.

99. Mr McDonald and Mr Robinson testified that, in monitoring the success of the company’s marketing strategy and the financial well-being of the company, the Sofology management did not separate out sales of Sofashield from its other sales, including the sales of sofas, in the figures for both total revenue and average order value. Instead, the Sofashield sales were included in those figures without being distinguished from those other sales.

100. On the other hand, the board packs which we were shown to us revealed that the board did monitor what it called “Value Added KPIs” – namely, the attainment of KPIs in relation to Sofashield, care kits and footstools. Mr McDonald confirmed that the attachment rate for each of those items was monitored on a weekly basis. The reason for this was that those attachment rates were closely linked to the average order value. For example, the May 2016 monthly board pack noted that the average order value for the year to date was behind both budget and the previous year in large part because of a “lower penetration of Sofashield and care kits”.

DFS

Contractual arrangements

Contractual arrangements with customers

101. During the DFS relevant period, Sofacare was not a product which was sold in isolation and a customer could not purchase Sofacare without purchasing a sofa. A customer could purchase Sofacare only:

at the same time as he or she purchased the relevant sofa; or

at any time up to delivery of the sofa.

(In these respects, Sofacare was unlike some of the other additional items sold by DFS along with sofas, such as occasional tables and footstools.) However, Sofacare was available on virtually all of the sofas sold by DFS during the DFS relevant period. The only exclusions were sofas previously returned by customers and sofas used on commercial premises.

Contractual arrangements with Guardsman

102. We were provided with the contract between DFS and Guardsman dated 1 June 2017 which was in force for the remainder of the DFS relevant period. Prior to that date, the company had two insurance intermediary contracts in place – one with Guardsman for fabric sofas and the other with Castelan for leather sofas. We were not provided with either of those contracts but we have proceeded on the basis (which Mr Smith confirmed to be reasonable) that each of the two contracts in relation to the insurance intermediary services which applied for the earlier part of the DFS relevant period broadly followed the same form as the contract we saw.

103. Under the terms of the agreement, DFS was required:

to promote Sofacare to its customers in connection with its sales of sofas, using its own retail brand or the Guardsman retail brand;

accurately to set out the benefits and exclusions of the policy at the point of sale so as to enable the customer to make an informed decision about whether or not to purchase Sofacare;

to ensure that each member of staff selling Sofacare was properly trained; and

to ensure that, before any sale of Sofacare could be made online, the customer had to opt in.

104. In addition:

the contract provided that Guardsman undertook to pay to DFS £350,000 in respect of marketing support for DFS to promote Guardsman’s policies and £150,000 in respect of the administrative costs which DFS might incur in switching to Guardsman’s policies. (At the hearing, we were informed that the company had been unable to locate any evidence that the above amounts had been paid and Mr Smith said that, whilst he believed the second amount to have been paid, he didn’t know whether the first amount had been paid. We can see no reason to think that DFS would have failed to pursue either of the two payment obligations given that they were due shortly after the execution of the contract and therefore we have concluded that, on the balance of probabilities, both sums were in fact paid); and

there was a significant difference between the amount charged by DFS to the customer and the amount payable to Guardsman by DFS with the result that, even after taking into account the fact that the former amount included insurance premium tax, DFS made a significant profit out of each policy which it sold.

Contractual arrangements in relation to PPC advertising on Google

105. Unlike Sofology, DFS did not have a direct contract with Google. Instead, it entered into a contract with an agency called MediaCom Holdings Limited (“MediaCom”) to manage the PPC advertising on its behalf. As part of its remit, MediaCom contracted with Google and managed the operation of DFS’s “shopping” and “search” adverts on Google. Google billed MediaCom and those charges were then on-charged by MediaCom to DFS.

106. We were not provided with a copy of the contract between MediaCom and Google. However, we think that, on the balance of probabilities, that contract would have contained the same terms as are described in paragraphs 66 and 67 above in relation to the contract between Sofology and Google. This is because the terms of the contract between Sofology and Google appeared to us to be “standard-form” and, hence, generic and not bespoke in nature.

Objectives

107. Mr Brewer said that the overall goal of DFS’s digital marketing was to build the DFS brand (as a leading sofa retailer) in relation to awareness of its sofa offering and the types of sofas which it offered by means of customer awareness of that sofa offering. In so doing, the company hoped to drive sales of sofas online, in-store and on the telephone. The goal was to get customers onto the website and to the most relevant page to meet their enquiry. To that end, he worked with MediaCom to set the keywords for each PPC advertising campaign, to monitor the success of the campaign on an ongoing basis and to adjust the keywords where appropriate.

108. He added that driving Sofacare revenue had never formed part of the marketing strategy. Sofacare was never used as an acquisitional tool for marketing to drive customers and was not a consideration when the company set its marketing strategy. Instead, the strategy was designed to increase sales of sofas. The aim was “to generate engagement with our brand to get customers to buy sofas and drive revenue on sales of the furniture products”.

109. This objective was reflected in the fact that:

it was not part of the digital strategy to drive visitors to the sections of the website which dealt with Sofacare; and

the company did not bid on any keywords which related to Sofacare so that typing in “Sofacare” as a search term did not give rise to a Sofacare advert as an impression on Google.

110. These views were echoed by Mr Ashworth, who said that:

the marketing strategy involved making the customer aware of its brand first and foremost and ensuring that the customer chose its brand over those of its competitors; and

Sofacare did not feature in any of the three classical communication pillars comprising the marketing strategy of the business – segmentation, targeting and positioning. Sofacare was not mapped or targeted and the attachment rate of Sofacare to sales of sofas was not a measure used by the marketing team to monitor its success. When the research company, Boxclever, were retained to conduct customer surveys, the surveys did not include any questions for customers in relation to Sofacare as it was not identified as a relevant consideration.

111. Mr Smith made the same point, when he said that:

“[our] marketing costs are treated as an investment to increase customer traffic to our showrooms and online, with the aim of increasing sales of our sofas…We do not consider that Sofacare drives a customer to buy a sofa from DFS on the basis that we offer insurance in the form of Sofacare as an add-on product; and

DFS was interested in monitoring the Sofacare attachment rate “to ensure that we see equal averages across the showrooms” and to ensure that sales of Sofacare were being conducted in compliance with the regulatory regime.

112. Having said that, Mr Brewer conceded that, as the KPIs included the aggregate value of sales and average order values and those figures included all additional items sold by DFS in addition to sofas, sales of Sofacare were examples of items which would increase those figures. In addition, we were shown:

an internal presentation in relation to the company’s high level search and performance strategy which stated that the objective was to drive high quality traffic online and into stores to achieve overall business sales and included KPIs such as unique website visitors, market share, the “click-through rate” (the numbers of clicks on adverts) and the “bounce rate” (occasions when a person clicks on an advert and enters the website but leaves the site immediately without further interaction); and

an example of a weekly report from MediaCom which contained similar KPIs,

and Mr Brewer accepted that the KPIs in relation to PPC advertising were measuring the number of users who clicked on an advert, the length of time spent online, market share, footfall in stores, aggregate sales and average order values and were not measuring sales of specific products.

Nature

113. The evidence of Mr Brewer was that, of DFS’s spending on PPC advertising during the DFS relevant period:

27.37% was spent on “shopping” adverts, generating 14.52% of the visitors to the DFS website from PPC advertising (the “paid traffic”);

14.10% was spent on brand “search” adverts, generating 41.26% of the paid traffic;

4.55% was spent on competitor “search” adverts, generating 1.53% of the paid traffic; and

53.97% was spent on generic “search” adverts, generating 42.68% of the paid traffic.

114. Mr Brewer went on to say that:

the breakdown of spending on PPC advertising fluctuated from time to time depending on the market, competitor activity and the available budget;

in general, the spending on generic “search” adverts was the highest because, “if we want to grow our share in the market, this has to be done through maximising the efficacy of our generic searches”; and

the spending on PPC advertising was allocated first to brand “search” adverts and then anything left over was allocated to generic “search” adverts. The brand “search” adverts were critical for defending the DFS brand from inroads by its competitors and to get prospective customers to the DFS website ahead of the websites of its competitors.

The journey through the website

115. Each of the “shopping” adverts contained photographs of sofas and each of the “search” adverts referred to sofas as part of the text. Neither category of advert referred to the availability of Sofacare.

116. As for the journey which occurred when a user clicked on an advert, Mr Brewer testified that:

approximately two-thirds of the brand “search” adverts led the user to the DFS home page as the landing page;

the majority of the competitor “search” adverts led the user to the DFS home page as the landing page; and

the majority of the generic “search” adverts led the user to the DFS home page as the landing page. The reason why so many generic “search” adverts went through to the home page as the landing page as opposed to a broad category page or an individual product page was because “we’re trying to establish the type of sofa that the individual wants before they’ve made their decision”.

117. Mr Vernon explained that the information about Sofacare on the website was accessible by way of a link from the page of the website dealing with caring for sofas generally. (Somewhat confusingly, Sofacare shares its name with the term used by DFS for caring for sofas in general and, to avoid confusion in this decision, we refer to the page of the website dealing with caring for sofas in general as the “Care page”.) Google analytics data for the DFS relevant period showed that the Care page was a landing page for 0.003% of visitors who had clicked on a “search” advert during that period but Mr Vernon said that he and his colleague, Ms Catherine Woodward, the senior digital marketing manager for DFS, thought that even this low figure was incorrect because they would not expect any visitor who reached the website by clicking on a “shopping” or “search” advert to be directed to that page. He ascribed this anomaly to the cookie-based Google analytics system pursuant to which a later visit to a specific page on the website other than through an advert might incorrectly be attributed to an earlier visit to the homepage through an advert.

118. The “shopping” adverts were different in that customers were presented with images of sofas and so clicking on one of the images would take the visitor directly to an individual product page.

119. Both the home page and each broad category page (as well as each individual product page) showed images of sofas and other items of furniture and made no reference to Sofacare. However, there was a navigational toolbar at the top of the relevant pages and one of the items on that toolbar was entitled “Help”. By holding the cursor over the word “Help”, the user was able to see a menu of miscellaneous items, one of which was the link to the Care page. The user could then click on that link to get to the Care page and that page then had links to three other care pages – one for fabric sofas, one for leather sofas and the third for bedroom and dining furniture. The fabric care and leather care pages both contained details of Sofacare, along with other information on how to care for sofas.

120. Thus, by clicking on a link on the Care page, the customer was taken to a page with further information on how to care for the relevant category of sofa and what Sofacare included or excluded. That page also included a promotional video dealing with caring for sofas and the video included information about Sofacare.

121. Mr Vernon said that the Care page tended to be looked at by customers who had previously purchased a sofa and wanted to learn about tips for caring for their sofas or whether or not the Sofacare which they had purchased covered them in respect of a particular incident of damage. This accounted for the relatively low number of times that the page was viewed, relative to overall pages viewed (0.09%). The content on Sofacare accessed through the Care page did not refer to pricing or provide information on how to purchase Sofacare. It merely set out the inclusions and exclusions. There was therefore no page on the website which was exclusively devoted to Sofacare as such, as distinct from other information on how to care for sofas.

122. In addition to the information on Sofacare accessed through the Care page, as described above, an option to add Sofacare and/or a care kit to the user’s basket appeared on the website once the relevant sofa had been added to the basket (apart from in the rare cases mentioned in paragraph 101 above) and the user had registered. It was the first thing that appeared once the registration process was completed. (The purpose of requiring users to register at an early stage was so that the company could subsequently send email chasers to a person who failed to complete a purchase following registration.) A promotional video in relation to caring for sofas – which was different from the video accessed through the Care page - could also be viewed at that stage.

123. Each individual product page contained details of other items of furniture in which the user might be interested, to which the user was able to navigate by clicking on the relevant item.

124. The Google Analytics data revealed that, within a one-year period falling within the DFS relevant period:

the average number of pages viewed by a visitor to the DFS website who had clicked on a Google advert was approximately 4.6 pages per visit;

the average number of pages viewed by a visitor to the DFS website who was visiting by way of a free search was 4.78 pages per visit for a visitor who had not previously visited the DFS website within the previous year and 4.58 pages per visit for a visitor who had done so;

of those visitors to the DFS website who had clicked on a Google advert, 0.30% of visitors who had not previously visited the DFS website within the previous year and 0.42% of visitors who had done so made online purchases in the course of the visit;

of those visitors to the DFS website who were visiting by way of a free search, 0.31% of visitors who had not previously visited the DFS website within the previous year and 0.40% of visitors who had done so made online purchases in the course of the visit.

125. It therefore appears that, over that one-year period, as between those people who visited the DFS website as a result of clicking on a Google advert and those people who visited the DFS website as a result of a free search, there was very little difference in both the numbers of pages viewed or the percentages of visitors making purchases.

Training

126. The sales induction workbook from July 2016 trained sales staff to sell all DFS products, including Sofacare, using the “REACH” technique. “REACH” was an acronym for “[Building] Relationship”, “Establishing [Needs]”, “Advice [and Reassurance]”, “Conclude” and “[End on a] High”.

127. Ms Johnson explained that staff were trained to the effect that:

as part of the “Establishing Needs” part of the “REACH” process, they should ask questions about the customer’s personal circumstances so that that information could be used not just to find the right sofa for the customer but also better to sell Sofacare to the customer once the customer had decided to purchase the sofa. (Ms Johnson initially said that the information was obtained solely in order to establish whether the customer needed Sofacare but she subsequently conceded that there were no circumstances in which Sofacare would not be offered to the purchaser of an eligible sofa;)

at the “Advice and Reassurance” part of the “REACH” process, they should promote the key benefits and features of the products which they wished to sell;

they should not raise the subject of Sofacare until after the customer had decided to purchase a sofa because raising it too early could lead to the loss of the sofa sale. This was because the customer would then start to price-in the additional cost of the Sofacare in deciding whether or not to purchase a particular sofa; and

conversely, they should raise the issue of Sofacare at the point when the customer had decided to proceed with making the sofa purchase.

128. We were shown a document dated November 2016 setting out the DFS policies and procedures guidelines for sales of Sofacare. The document was in the form of a flow chart which was colour coded to match the five different stages in the “REACH” process. This showed that, during the “Advice and Reassurance” stage, if the customer initially declined to take up Sofacare, the relevant sales-person was trained to discuss the reasons for the customer’s objection and try to persuade the customer to take it up by offering it again. The document went on to remind sales staff that no undue pressure should be placed on the customer to purchase Sofacare and that the company considered that more than two attempts to sell Sofacare to the customer would not be best practice.

129. Sofacare was not the only additional item in respect of which sales staff received training. For example, we were provided with an example of training materials which were designed to teach staff how to maximise sales of footstools.

Selling methods

130. As regards sales online, the customer was given the option to add Sofacare to the virtual basket only after first placing the sofa in the virtual basket and registering in the course of the checkout process. At that stage, the option to add Sofacare and/or a care kit was the first option to be offered.

131. Mr Ashworth explained that, based on surveys conducted for DFS by a research agency, Boxclever, the company was aware that the average customer would take three weeks from the time that he or she decided to buy a new sofa until completing that purchase and would make two store visits in that time.

132. One of the objectives of DFS’s PPC advertising during the DFS relevant period was to drive footfall into stores. As regards such sales, Ms Johnson said that:

certain customers came to a store with a particular sofa in mind and did not want to look at anything else, whereas others were either just browsing or had a particular sofa in mind but were interested in trying out other sofas too;

the sales staff used the “REACH” process as the framework for interacting with customers;

although sales staff were instructed not to raise the issue of Sofacare prior to the time when the customer had made a decision to buy the relevant sofa, if the customer raised the subject himself or herself, then the sales-person would deal with the query before returning to the sofa sale;

at the point when Sofacare was raised with the customer, after the customer had decided to proceed with the sofa purchase, the customer was given a Sofacare leaflet and informed of the cost of buying Sofacare;

the sofa, together with the Sofacare and any other additional items, were then invoiced at the same time; and

the sales staff were incentivised to sell Sofacare by the fact that they earned commission on sales of Sofacare in the same way as they earned commission on sales of sofas. However:

the sales staff were instructed that, if they put inappropriate pressure on customers to buy Sofacare, that could result in disciplinary action and loss of incentives; and

attachment rates were monitored and, if a sales-person’s attachment rates were noticeably higher (or lower, for that matter) than the average, there was a system in place for investigating that.

133. As regards sales by telephone, we were shown the script used by the DFS operators from July 2019. This revealed that, after the customer had agreed to buy the sofa, operators took steps to sell Sofacare on the basis of the information previously revealed by the customer during the sales process in relation to the sofa.

Financial information

134. Over the DFS relevant period:

the respective attachment rates were 73% for in-store sales, 24% for online sales and 37% for telephone sales; and

the attachment rate for the business as a whole was 62%.

135. Mr Smith said that attachment rates for each category and the business as a whole tended to remain fairly constant over time (although there tended to be a dip in the in-store attachment rates in the run up to Christmas when the in-store sales staff were stretched and additional assistants who were not trained to sell Sofacare had to be drafted in.) The attachment rate for the business as a whole tended to be around 60%.

136. Sofacare made a significant contribution to the company’s gross sales and gross profit over the DFS relevant period. Sofacare accounted for 7.3% of the company’s gross sales and 12.7% of the company’s gross profit. The fact that the percentage of gross profit was so much higher than the percentage of gross sales reflected the greater profit margin implicit in sales of Sofacare.

137. The company held a weekly meeting at which a trading pack containing information relating to the previous week and the year to date was discussed.

138. Mr Smith took us through an example trading pack which showed that, in monitoring the financial well-being of the company:

the DFS management did not separate out sales of Sofacare from its other sales, including the sales of sofas, in the figures for daily net orders and average order value. Instead, the Sofacare sales, as with other additional items such as coffee tables, footstools or an upgrade in the sofa leather, were included in those figures without being distinguished from those other sales;

average order value was an important KPI and persuading customers to purchase Sofacare (or any other additional items) would increase average order value;

the fact that the average order value for online sales was lower than the average order value for in-store sales was attributable in part to the lower attachment rate for online sales and in part to the fact that online customers tended to purchase less expensive products; and

sales of Sofacare, as well as other additional items such as care kits and footstool, were KPIs which needed to be tracked because of their impact on the daily net orders and the average order value.

139. The example trading pack showed that, over the 10-week period to which the pack related, sales of insurance amounted to a little over half of the VAT-exclusive gross revenue which had been generated in respect of the additional items as a whole (£13,020,000 out of £25,627,000).

140. Sales of Sofacare were also included as an indistinguishable part of revenue in both the interim accounts dated 9 March 2021 and the budget for the 2018 financial year dated 5 September 2017. Mr Smith agreed that the latter demonstrated that the marketing budget was a function of the aggregate revenue of the company, with the result that a reduction in sales of Sofacare would have a depreciatory effect on the marketing budget.

141. Finally, we were taken to:

various viability statements which were prepared from time to time by the company for its auditors. These demonstrated the potential impact of certain risks on the ability of the company to meet its financial covenants to its lenders and identified the steps which might be taken to mitigate any of the risks which, if it materialised, would pose a threat to the company’s compliance with those covenants. The viability statements identified that one of the potential risks for the company was a decline in the sales of Sofacare as a result of a more robust regulatory environment and concluded that, in certain scenarios, no mitigation steps would be required to maintain compliance with the financial covenants but that, in others, mitigation steps would be required. One of the mitigation steps listed was a reduction in the marketing budget; and

a budget update dated June 2019 which mentioned the risk of potential changes in the regulatory environment as a result of an ongoing review by the Financial Conduct Authority as one of the macroeconomic risks facing the company and calculated that, after applying a risk-weighting to each of the macroeconomic risks facing the company, those potential regulatory changes posed the greatest risk to the company in terms of a reduction in its profits before tax.

findings of fact

142. Based on the evidence described in paragraphs 63 to 141 above, we make the following relevant findings of fact:

the purpose of the relevant Appellant in purchasing PPC advertising was to encourage prospective customers:

to enter its website, as opposed to the websites of its competitors; and

to go into its stores, as opposed to the stores of its competitors

in each case, for the purpose of acquiring sofas;

we have reached that conclusion because:

that was each Appellant’s marketing strategy in purchasing PPC advertising. The strategy was directed at increasing sales of sofas and building awareness of the relevant Appellant’s sofa offering – see paragraphs 68, 69 and 107 to 111 above;

each of the “shopping” adverts contained photographs of sofas and each of the “search” adverts referred to sofas as part of the text – see paragraphs 79 and 115 above; and

when clicking on a “shopping” or “search” advert, the user was taken to a landing page which showed images of sofas – see paragraphs 81, 116 and 119 above;

conversely, neither Appellant purchased PPC advertising with the purpose of encouraging prospective customers to purchase insurance;

we have reached that conclusion because:

neither Appellant’s PPC advertising strategy was directed specifically at increasing sales of insurance or was driven by the relevant Appellant’s insurance offering – see paragraphs 68, 69 and 107 to 111 above;

the fact that the Appellant offered insurance was not a feature which the Appellant sought to emphasise in its “shopping” or “search” adverts. Neither Appellant made bids in the Google auctions for words relating to insurance and insurance did not feature as a search term in any of the adverts– see paragraphs 69, 79, 108, 109, 110(2), 111(1) and 115 above;

a user clicking on a “shopping” or “search” advert was never taken to a landing page that highlighted the relevant Appellant’s insurance offering. That offering was accessible from the landing page by hovering on the “More” or “Help” button on the navigational toolbar at the top of the landing page and then scrolling down to the relevant link but it was not something with which the user was immediately confronted when opening the landing page and it did not feature prominently on that page – see paragraphs 81, 82, 117, 119 and 120 above;

unless the user chose to navigate to the page which dealt with the relevant Appellant’s insurance offering, insurance was not part of the user’s journey on the website until after the sofa had been added to the user’s basket – see paragraphs 83 and 122 above; and

it was not possible for a prospective customer to purchase insurance without purchasing a sofa, whether online, on the telephone or in-store. A purchase of insurance could be made only at the same time as he or she purchased the relevant sofa or at any time up to delivery of the sofa – see paragraphs 63 and 101 above;

the respective significance to each Appellant of its sofa sales relative to its insurance sales could be seen in the fact that sales staff were trained that, unless the subject of insurance was raised by the prospective customer himself or herself at an earlier stage, insurance was not to be raised with the prospective customer until he or she had decided to purchase a sofa, for fear of losing the sofa sale – see paragraphs 89, 127(3) and 132(3) above;

on the other hand, given its contractual obligations to the insurer for which it was acting as intermediary but, perhaps more particularly, given the high margin which it earned on its own account from arranging sales of insurance, each Appellant was keen to ensure that as many of its sofa sales as possible were accompanied by sales of insurance. The evidence we heard showed that the insurance attachment rate was highly significant to each Appellant from the commercial perspective because it increased both the relevant Appellant’s profitability and the relevant Appellant’s average order value. It was a highly profitable business stream in its own right – see paragraphs 98, 136, 138(2) to 138(4) and 139 above. As such, each Appellant took steps to ensure that as many of its sofa sales as possible were accompanied by sales of insurance. For instance:

in each case, sales staff were trained to use information provided by the customer during the negotiations in relation to the sale of the sofa to push the insurance and to ensure, so far as allowed by the prevailing regulatory regime from time to time, that customers were encouraged to add the insurance to their sofa purchase – see paragraphs 85 to 88, 90 to 92, 94, 126 to 128, 130, 132(4) and 133 above;

for each Appellant, the attachment rates were high and the attachment rate for in-store sales was higher than the attachment rates for online or telephone sales, reflecting the power of face-to-face personal persuasion on the part of the sales staff. The fact that there was a dip in the DFS in-store attachment rates in the run up to Christmas when the in-store sales staff were stretched and additional assistants who were not trained to sell Sofacare had to be drafted in also evidenced that fact – see paragraphs 96, 134 and 135 above;

each Appellant’s marketing strategy involved driving footfall into stores, where the attachment rates were higher – see paragraphs 72, 73 and 132 above;

each Appellant monitored attachment rates as one of its KPIs – see paragraphs 100, 137, 138(2) to 138(4) and 141 above;

in the case of Sofology, the company operated an “opt-out” basis in relation to sales of insurance. In other words, a customer purchasing a sofa was told that the insurance had been added automatically and had to ask for the insurance to be removed – see paragraphs 86, 87 and 94 above.

There was a dispute between the parties at the hearing in relation to how long the “opt-out” practice continued during the Sofology relevant period. The training materials referred to in paragraph 86 above showed that it was definitely part of the practice of the company in January 2016 and that it had ceased by December 2016 but we were provided with no evidence by Sofology to the effect that it did not operate within the Sofology relevant period prior to January 2016 or that it ceased any earlier than December 2016. In addition, the telephone script referred to in paragraphs 87 and 94 above suggested that the “opt-out” basis applied to telephone sales throughout the Sofology relevant period. Given:

the significance of this point; and

the fact that, despite being raised by the Respondents in their skeleton argument before the hearing, Sofology produced no evidence to suggest that the practice did not operate prior to January 2016 or ceased any earlier than December 2016,

we consider that Sofology has failed to discharge the burden of proof on this point. We accordingly find as a fact that the practice of requiring customers to opt out instead of opting in in relation to the insurance continued throughout the Sofology relevant period.

The same was not true of DFS in the DFS relevant period. We have already observed that the contract between DFS and Guardsman expressly required that sales of insurance online needed to be made on an “opt-in” basis. There was no equivalent to that clause in the clause in the contract dealing with in-store sales but we have seen no evidence to suggest that DFS adopted an “opt-out” basis in relation to any of its sales. Moreover, the DFS relevant period fell somewhat later than the Sofology relevant period, at a time when the regulatory climate was likely to have been stricter and an “opt-out” basis would have been prohibited. Accordingly, we find as a fact that DFS operated an “opt-in” process in relation to insurance throughout the DFS relevant period;

in the case of Sofology, the company took steps to ensure that there was a close connection between its brand and its insurance offering. For example, Sofology used the brand strapline “Feel at home on the sofa you love” both in relation to its sofas and in relation to Sofashield – see paragraph 71 above;

in the case of DFS, staff were incentivised to sell insurance by being paid commissions which took the sales of insurance into account – see paragraph 132(6) above; and

each Appellant produced leaflets and/or videos extolling the benefits of the insurance – see paragraphs 81, 92, 120, 122 and 132(4) above;

in addition, when it came to calculating revenue, profitability and average order value, each Appellant did not distinguish the income which it derived from sales of insurance from the income it derived from sales of sofas– see paragraphs 72, 73, 99, 112, 138(1) and 138(2) above - and hence:

for each Appellant, a reduction in sales of insurance would have reduced the relevant Appellant’s marketing budget as a whole – see paragraphs 97 and 140 above; and

DFS took the attachment rates into account in assessing its ability to meet its financial covenants and its macroeconomic risks – see paragraph 141 above; and

finally, it is apparent from the evidence in relation to the landing pages, the website layout and, in the case of DFS, the Google analytics data, that a user entering into the relevant Appellant’s website by way of clicking on a Google advert was encouraged to look at a variety of different sofas or other items of furniture offered by the relevant Appellant and not simply to look at the specific sofa which had led the user to click on the “shopping” or “search” advert in the first place. As such, clicking on an advert during the relevant period might well have led to the purchase of a sofa or other item of furniture which was different from the sofa to which reference was made in the advert – see paragraphs 81, 82, 84, 116(3), 119 and 123 to 125 above.

For completeness, we should record that we were not provided with the same level of Google Analytics data in relation to Sofology over the Sofology relevant period as we were in relation to DFS over the DFS relevant period. For example, we were not provided with information about the number of pages visited by the various different categories of visitor on each occasion that a person visited the Sofology website during the Sofology relevant period. It may well be the case that that information would have been very little different from the information set out in paragraphs 124 and 125 above in relation to DFS but it could equally well have been very different and so we are unable to make any finding of fact in that regard. In any event, as we have already noted, Sofology did not make online supplies of sofas until the final month of the Sofology relevant period.

the arguments of the parties

Introduction

143. Ms Sloane accepted that, in order for each Appellant to succeed in its appeal, it would need to show that the PPC advertising costs which it had incurred during its relevant period:

were not overheads; and

had a direct and immediate link solely to its taxable supplies of sofas and not to its exempt supplies of insurance intermediary services.

She submitted that both of those were the case and that, in the case of each Appellant, the advertising costs had a direct and immediate link solely to the relevant Appellant’s taxable supplies of sofas.

The primary argument

144. The primary argument to the contrary advocated by Ms Barnes and Mr Reynolds was that there was a direct and immediate link between the costs of the Google adverts and both the standard-rated supplies of sofas and the exempt supplies of insurance intermediary services.

The nature of the test – the relevance of subjective intention

145. Their starting point was that the authorities made it clear that this was a multi-factorial test which was highly fact-specific and objective – as was made clear in Midland Bank at paragraph [25]. They said that, in weighing up the relevant factors and conducting that objective examination, the subjective intentions of the relevant Appellant were one of the factors to be taken into account, albeit that the conclusion was not wholly dependent on those subjective intentions. This was not saying that one needed to look at the taxable person’s ultimate aim. They accepted that BLP had made it clear that that was not the test. It was more that the subjective intentions of the taxable person were part of the overall factual matrix.

146. Ms Sloane disagreed with that approach. She said that, when the CJEU in Midland Bank held, at paragraph [25], that the national court had to consider all the facts and circumstances in determining whether there was a direct and immediate link between an input and an output, it was merely saying that the issue was highly fact-dependent and there was no “one size fits all” way of expressing the test. It was not saying that the relevant test was a free-for-all in which every factor or circumstance was potentially relevant.

147. In her view, contrary to the submissions of Ms Barnes and Mr Reynolds, the subjective intentions of the taxable person were wholly irrelevant. She pointed out that, in ROH, David Richards LJ had said the following at paragraph [17]:

“The taxable person’s purpose is to be objectively ascertained from the facts and circumstances of the transactions, not by investigating the subjective intentions of the taxable person”.

148. Related to this point was the fact that, as noted in BLP, the ultimate aim of the taxable person was wholly irrelevant – see BLP at paragraph [19] and ROH at paragraphs [49] and [59]. To take the taxable person’s ultimate aim into account would be “contrary to the VAT system’s objectives of ensuring legal certainty and facilitating application of the tax by having regard, save in exceptional cases, to the objective character of the transaction in question” (see BLP at paragraph [24]). As had been noted by Advocate General Kokott in ‘Sveda’ UAB v Valstybiné mokesčiu inspekcija prie Lietuvos Respublikos finansu ministerijos (Case C-126/14) STC 2016 447 (“Sveda”) at paragraph [45], there needed to be an objective economic link between the input transaction and the output transaction and the input transaction had objectively to serve the purpose of the output transaction.

149. It followed from this that the subjective intentions of the personnel at each Appellant were wholly irrelevant. An advert relating directly and immediately to the supply of a specific category of goods or services did not become an overhead or become directly and immediately related to the supply of another category of goods or services merely because of the way that the taxable person in question viewed the benefits of the advert or measured the results of the advert – see ROH at paragraph [17].

150. Ms Sloane added that, even if the taxable person’s subjective intentions were to be a relevant part of the factual matrix, the fact that the taxable person was interested in identifying and quantifying all of the benefits which flowed from a particular type of expenditure said nothing about whether or not that type of expenditure had a direct and immediate link to a particular category of supply which represented one of those benefits. This was because a business was interested in identifying all of the benefits resulting from the expenditure which it incurred, regardless of whether those benefits were direct and immediate or were instead indirect. The business was solely interested in results, however those results arose.

The nature of the test – physical content

151. Ms Sloane said that what was relevant in identifying the supplies to which the cost of an advert had a direct and immediate link was primarily the content of the advert and, examined in the light of that content, the content and layout of the landing page. This was because the relevant issue was to identify the goods or services to which the advert related and to which the advert was directly attracting the prospective customer. That approach was entirely consistent with the case law on advertising - see Skipton and DaP – and it was why the Court of Appeal in ROH said at paragraph [46] that the link between the production costs and the catering supplies in that case was not analogous to the link which would have existed between an advert for the catering and the catering supplies.

152. Ms Barnes and Mr Reynolds said that, on the contrary, little could be drawn from the acceptance by the Court of Appeal in ROH at paragraph [46] that there would be a direct and immediate link between the costs of an advertising campaign for the catering and the subsequent supplies of the catering because that was merely an obvious example of a direct and immediate link in the advertising context. It did not mean that advertising costs could have a direct and immediate link to the catering only if the advertising campaign related to the catering.

153. They added that Ms Sloane’s reliance primarily on the content of the adverts, and, in the light of that content, the content and layout of the landing pages, as being determinative was overly-reductionist and flawed. There were four problems with Ms Sloane’s approach, as follows:

it did not take into account all of the facts and circumstances surrounding the adverts and the Appellants’ supplies, as was required by, inter alia, ROH at paragraph [17]. It was clear from the proliferation of case law in this area that this was a complicated question. That would not be the case if the answer could be found solely by looking at the physical content of the advert and the content and layout of the landing pages. Moreover, it was a blinkered approach - which was the precise opposite of taking into account all the facts and circumstances;

focusing on physical content led to absurdity. Ms Sloane had accepted that an advert for a specific type of sofa could be seen as having a direct and immediate link to supplies of sofas more generally and not just to the specific sofa to which the advert referred. But then where did that stop? Could the link be extended to other products offered by the Appellants, such as insurance? In addition, if an item featured in any way in an advert or on the landing page, was that sufficient to create a direct and immediate link between the advertising cost and the relevant item or did one need to consider the prominence of the item in the advert or on the landing page? The latter would be highly subjective and the precise opposite of the objective approach required by the relevant case law;

the focus on physical content was not compatible with the manner in which PPC advertising worked. With PPC advertising, the retailer incurred the cost of the service only when the user clicked on the advert. The mere impression of the advert on the Google page incurred no charge. Thus, the retailer was paying not for the advert per se but rather for the re-direction of the user to the retailer’s website as a result of clicking on the advert. This meant that the physical content had necessarily to include the content of the landing page on the retailer’s website. This was different from other forms of advertising. And, once one took into account the content of the landing page in this case, one had to recognise that insurance was mentioned on that landing page through the drop-down menu on the navigational toolbar; and

focusing on physical content alone required certain facts which were plainly relevant in answering the question to be ignored. For example, it required facts like the attachment rates and whether or not the insurance was “opt-in” or “opt-out” to be disregarded and that could not conceivably be right.

154. They said that, notwithstanding the above, in Skipton, the First-tier Tribunal had based its decision solely on the physical content of the adverts. The decision should therefore be approached with caution, given the rest of the case law. However, even if one were to apply the reasoning in Skipton in the present case, the fact that the insurance was mentioned in the drop-down menu off the navigational toolbar meant that the facts in the present case could be equated to those adverts in Skipton which mentioned the mortgages and that approach would be sufficient to establish a direct and immediate link between the cost of the adverts and the supplies of insurance intermediary services.

155. In response, Ms Sloane did not demur from the proposition that the content and layout of the landing pages were factors which needed to be taken into account, in addition to the content of the adverts, but she said that:

she did not agree with the proposition that the content of the adverts (the impression on Google) could simply be disregarded. She pointed out that the Respondents’ approach in this respect demonstrated a confusion on their part between the nature of the service which Google was providing and the circumstances in which a payment for those services was triggered. It was quite clear from the contractual arrangements with Google that the service that Google was providing was that of a PPC advertising platform and therefore the content of the adverts was of greater significance in answering the question than the Respondents were accepting. It was that content which was chosen to entice the users to click on the adverts; and

the content and layout of the landing pages needed to be considered very much in the context of the content of the adverts, and not in isolation. The fact that a landing page might contain a reference, somewhere in its layout, to some goods or services which were unrelated to the goods or services to which the relevant advert referred would not generally create a direct and immediate link between the advert and those other goods or services. Every website page had links to other content (in the same way that a person going into a store would be able to access goods other than those which he or she had been induced by an advert to enter into the store to purchase). Of course, it was possible to think of extreme examples along the lines of the sprat and mackerel example used by the First-tier Tribunal in DFS1 but that extreme example proved the rule. If clicking on an advert for sofas took the user directly to the insurance page as the landing page, then that might well establish that there was a direct and immediate link between the cost of the advert and the supplies of insurance intermediary services. However, that was not the case here. In this case, the content and layout of the landing pages were entirely appropriate in the context of the content of the adverts. The adverts related to sofas and the content and layout of the landing pages were those which were considered by the relevant Appellant to be most appropriate for selling sofas.

156. She added that there were a number of problems with the Respondents’ approach of focusing so heavily on the all-embracing content of the landing page. These were that:

in many cases, the landing page was not the home page but was in fact a general category page or a specific product page. It would be odd if the answer were to differ depending on which particular landing page it was;

focusing solely on the content of the landing page involved ignoring the terms of the advert which had induced the user to click on the advert and enter the landing page;

the approach ignored the fact that the relevant Appellant’s home page operated in the same way as a store. In other words, the fact that it was possible to navigate away from the landing page to any of the relevant Appellant’s products was no different from the ability of a visitor to a store to ignore the first display to which he or she came and to explore the contents of the store;

when one took the contents of the landing pages into account in this case, there was no promotion of insurance on any of those pages. It was possible to navigate away from the landing pages to a page which dealt with the insurance but that was no different from the ability to navigate away from the landing pages to any product offered by the relevant Appellant in the virtual store. There was no promotion of insurance, as such, on any of the landing pages; and

the Respondents’ submission that the aim of the PPC advertising was to get the user onto the relevant Appellant’s website or into the relevant Appellant’s store was a point which could be made of all advertising in a general sense. But, in the case of all advertising, it was important to consider more specifically the underlying purpose for which the relevant retailer wanted to attract people onto the website or into the store. In this case, the purpose of each Appellant was to get the user onto the website or into the store for the specific purpose of purchasing a sofa. The fact that, once the user had decided to purchase a sofa, he or she was always offered add-ons did not mean that there was a direct and immediate link between the advert and each of the add-ons.

What each advert had done directly was to stimulate demand for the sofa and then the sale of the sofa had created the opportunity for the sale of the relevant add-on. This was as true in relation to the insurance in this case as it was for a shoe retailer who sold protector spray or a travel operator who sold holiday insurance. Just because there was a connection between the supplies of two different products did not mean that a cost which had a direct and immediate link with the supply of the first product also had a direct and immediate link with the supply of the second.

Another example was the case of an advert by Marks and Spencer for birthday cake. If a customer went into the store to buy a birthday cake as a result of the advert and then bought one of the other goods which were being offered in the store, there was a direct and immediate link to the supply of the birthday cake but the link to the supply of the other item would be indirect. It was not correct to conflate two distinct supplies because of their close economic links and thus conclude that the costs which had a direct and immediate link with one of the supplies necessarily had a direct and immediate link with the other – both ROH and Southern Primary were examples of this point (see ROH at paragraphs [85] to [88] and Southern Primary at paragraphs [25], [26], [32] and [33]).

The nature of the test - economic use

157. Ms Barnes and Mr Reynolds said that, once all of the facts and circumstances were taken into account, it was apparent that, objectively speaking, each Appellant used the adverts for the purposes of making its supplies of insurance intermediary services as well as making its supplies of sofas. This was more than saying that the supplies of insurance intermediary services would not have been made “but for” the adverts. It was a case of examining the relevant facts and circumstances objectively. The decision of the Court of Appeal in London Clubs made it clear that, in determining the extent to which input tax could be attributed to a supply in the light of the factual matrix, consideration was not limited to physical use but should take into account economic use. Etherton LJ had said the following at paragraph [34] in that case:

“The assessment must be of the real economic use of the asset, that is to say having regard to economic reality, in the light of the observable terms and features of the taxpayer’s business.”

158. Although they accepted that London Clubs concerned the apportionment stage of the process of determining input tax recovery – which is to say, the second stage of that process, that followed the initial stage of the process, of determining the supplies to which the input tax was attributable – the above principle was just as relevant to the initial attribution stage as it was to the subsequent apportionment stage. It was apparent from Lok’nStore at paragraph [40] that Henderson J considered that economic use was a relevant factor at both stages in the process. In that paragraph, Henderson J had counselled against keeping the two stages rigidly distinct from each other and said that, depending on the precise facts, considerations which were relevant at the attribution stage might also be relevant at the apportionment stage.

159. Ms Sloane responded that the case law relating to apportionment of input tax by reference to broad economic use was of no relevance when it came to the, logically-prior, question of identifying the supplies to which the relevant costs had a direct and immediate link. The extent to which each category of supply made economic use of input costs was clearly relevant when it came to apportioning those costs following the establishment of a direct and immediate link between those costs and each category of supply but economic use was not sufficient in and of itself to create that direct and immediate link. Thus, the decisions in London Clubs and St Helen’s were of no moment in this context because, in both of those cases, the existence of the direct and immediate link to both categories of supply had already been established before the dispute arose. The fact that the two stages of the process – attribution and then apportionment – should be kept distinct was shown in Lok’nStore at paragraph [43]. In that paragraph, Henderson J had made it plain that the distinction between the two stages in the process of determining input tax recovery should never be blurred.

160. In relation to the economic use of the advertising costs in this case, Ms Barnes and Mr Reynolds said as follows:

the income which each Appellant derived from its supplies of insurance intermediary services was significant and the decisions in each of DaP, DFS1 and N Brown showed that whether or not an advert gave rise to a meaningful revenue stream from a particular type of supply was a significant factor in determining the existence of a direct and immediate link between the cost of the advert and that type of supply. In that respect, there was a difference between the application of the test in the case of production costs and the application of the test in the case of advertising costs. When considering the former, where the costs had no inherent promotional value, it was appropriate to pay less regard to the economic consequences of the incurring of the costs than it was in the case of the latter, where the costs were being incurred to generate sales;

that was why they agreed with the approach adopted by the First-tier Tribunal in N Brown, where, in addition to considering the physical content of the adverts, the First-tier Tribunal had taken into account, as relevant (albeit not determinative) factors, the subjective intention of the individuals acting for the taxable person in choosing the content of the adverts, the response of prospective customers to the adverts and the ways in which the taxable person chose to measure the success of its marketing or set its marketing budget. Importantly, the First-tier Tribunal in that case had relied in part in reaching its conclusion on the fact that there was a predictable correlation between the sales of goods and the sales of credit and that therefore there was no need for the appellant to monitor the take-up of credit. That was similar to the present case where attachment rates for both Appellants were predictable;

the decision in DaP was important in various respects, as follows:

first, that case also involved input tax on advertising costs and showed that the court’s analysis should not start and end merely by considering the physical content of the relevant advert. For instance, a critical reason for the decision in that case was the financial significance of the insurance income to the taxable person’s business (and that significance was actually lower than the significance of the insurance income to each Appellant in this case). In addition, the Court of Appeal in that case considered that it was irrelevant that the advert did not refer to the supplies of insurance which would ultimately be made by the insurer after the three-month period of free insurance (which were the supplies that would generate commissions for the appellant). The adverts in that case had referred only to the three-month period of free insurance and that period did not give rise to any supplies by the appellant because it received no commissions for that period. The key factual findings by the First-tier Tribunal in that case were that:

the offer of free insurance was not intended merely to attract customers to the taxable supplies of airtime service contracts made by the appellant but was also intended to attract customers to stay with the insurer once the free period was over; and

the appellant had a direct financial interest in that being the case;

secondly, DaP was a case where the advertising costs were held to have a dual function and that was sufficient to result in the denial of part of the input tax incurred; and

thirdly, the Court of Appeal in DaP rejected the suggestion in that case that its approach involved either the improper application of a “but for” test or improperly taking into account the ultimate aim of the appellant. Instead, it approved of the conclusion reached by the First-tier Tribunal in that case to the effect that there was a direct and immediate link between the advert, with its reference to the free insurance period, and the eventual supply of exempt insurance intermediary services by the appellant once the free insurance period was over;

the decision in Mayflower showed that there was no direct and immediate link between production costs and supplies of catering made to the people who attended the ensuing performances of those productions and the decision in ROH showed that that was the case even where the supplies of catering were thoroughly integrated with those performances. In each case, the ratio of the relevant decision was that there was no direct and immediate link between, on the one hand, production costs and, on the other hand, the opportunity to earn revenue out of supplies of catering to the people who bought tickets to the productions which resulted from those production costs. Instead, that opportunity arose directly out of the exempt supplies of tickets to the performances. This contrasted with the link between the production costs and the supplies of programmes in Mayflower, where the fact that the content of the programmes made use of the productions was sufficient to establish the direct and immediate link. Each of these conclusions was unsurprising in the light of the prior case law;

the decision in ROH did not herald a new dawn but was simply a straightforward application of the principles established in the prior case law. There were fundamental and critical differences between the production costs in ROH and the PPC advertising costs in this case. In ROH, the production costs led to the ticket sales and so were directly “promotional” of the ticket sales but they were not directly “promotional” of the catering supplies. Instead, what led to the opportunity to make the catering supplies was the ticket sales. It was the ticket sales which brought the customers to the performances and thus it was they and not the production costs which were “promotional” of the catering supplies. In contrast, in this case, the PPC advertising costs were “promotional” of both the supplies of sofas and the supplies of insurance intermediary services. Thus, there was a direct and immediate link between the PPC advertising costs and the supplies of insurance intermediary services which was distinguishable from the indirect link between the production costs and the catering supplies in ROH. In short, each Appellant was using the PPC advertising costs to make its supplies of insurance intermediary services and the former were accordingly cost components of the latter;

the decision in ROH merely reiterated that dual consequences were not enough, in and of themselves, to establish a direct and immediate link between the relevant costs and both categories of supply. It did not establish that there could never be a direct and immediate link between the relevant costs and both categories of supply, as the earlier decision in DaP had shown. In ROH, the First-tier Tribunal had found, as a fact, that investment in the productions attracted customers not only for the performances but also for the catering but the Court of Appeal held that that did not mean that there was a direct and immediate link between the production costs and the catering because that was merely an economic consequence. It was the “but for” test couched in terms of a test of economic necessity. The costs were not being used to provide the catering. For instance, a payment for a costume or a payment to an artist was not being used to make a supply of champagne. The parallel in this case would be an argument that the cost of the materials used to make the sofas had a direct and immediate link to the supplies of insurance intermediary services. The PPC advertising costs in this case were not on all fours with this example. Instead, those costs were used to sell both sofas and insurance;

the facts in RAC were very different from those in this case. In this case, the supplies of insurance intermediary services were made at the same time as the supplies of sofas and there were no intervening chain-breaking exempt supplies between the incurring of the costs and the supplies of insurance intermediary services;

in DFS1, the First-tier Tribunal had considered the content of the adverts to be relevant but not determinative. It had looked at the role played by the advert in the business so that, for example, an advert for a sprat which was designed to sell a mackerel would have a direct and immediate link with the sale of the mackerel. They agreed with this approach. In addition, the First-tier Tribunal in DFS1 had rightly taken into account certain factors which it said were relevant to this question, one of which was the commercial importance of insurance to DFS and the attachment rates. However, the facts in that case were distinguishable from the facts in the present case. The insurance wasn’t as widely available on sofas at that time – insurance was available on only 38% of sofas - and the case concerned conventional advertising and not PPC advertising, which operated very differently, as noted in paragraph 153(3) above; and

the PPC advertising was being used to persuade the user to enter the relevant Appellant’s website or to go to a store and not to promote a specific product. It was ultimately more about the promotion of the relevant Appellant’s brand than about the promotion of a specific product – that was clear from the witness evidence, from the fact that the relevant Appellant’s home page or a broad category page (and not a specific product page) was the most common landing page and, in the case of DFS, from the Google analytics data showing the number of pages visited by, and the purchases made by, the people who clicked on the adverts (where the relevant figures were similar to the equivalent figures for people who entered the website through an organic search). The decision in Britannia was an example of a case where the physical content of the advertising was held not to be determinative because the advertising was designed to promote the appellant’s brand as a whole. The same was true in this case.

161. In response, Ms Sloane said that:

the importance of physical use in determining whether or not a direct and immediate link existed could be seen in Mayflower at paragraph [43], where the fact that the productions to which the costs related provided the subject matter of the programmes and “were as much part of the raw material used in preparing the programmes, as the paper and ink from which they were physically made” was sufficient to establish the direct and immediate link;

although that case and ROH - where the Upper Tribunal again adverted to physical use when referring to the direct and immediate link between the production costs and the ticket sales - both related to production costs and not advertising, it was not the case that, with production costs, one could take into account physical use but, with advertising costs, one could not and had to focus instead on economic use. The nature of the relevant cost undoubtedly affected the answer to the question of whether or not there was a direct and immediate link between that cost and a particular supply but the nature of the test was the same;

that was not to say that the issue of economic use was irrelevant. Economic use was a relevant factor but the key point was that the economic use had to be direct and immediate. Indirect economic use, no matter how essential to the business, was insufficient. The Respondents’ case in these appeals was that a plethora of indirect, albeit commercially-substantial, economic links were sufficient to comprise a direct and immediate link and that argument had been dismissed by the Upper Tribunal and the Court of Appeal in ROH (see ROH at paragraphs [46] and [47]). A cost could be commercially necessary in order for a supply to be made and economically used in making the supply but, if its use in making the supply was indirect, there would not be a direct and immediate link between it and the supply;

in this case, there was no such direct and immediate economic use. Neither of the Appellants had made any bids in the Google auctions in respect of keywords relating to its insurance products, none of either Appellant’s adverts referred to insurance and no user clicking on either Appellant’s adverts was taken to a landing page featuring the availability of insurance. There was thus no evidence that any of the adverts was using the availability of insurance as a lever or incentive or that any user was induced to click on an advert by reason of the availability of the insurance. Indeed, it was a matter of common sense that no-one would choose one sofa retailer over another because of their respective insurance offerings;

commercial and economic links, no matter how substantial, did not convert an indirect link into a direct and immediate link. As noted by the Court of Appeal in ROH at paragraph [100]:

“It is not enough to express the ‘but for’ test in economic terms and then contend that the link must be considered to be direct and immediate”.

Thus, the importance of the insurance intermediary services as a revenue stream for the relevant Appellant, whether the income from those services subsidised the relevant Appellant’s marketing budget and whether the adverts were commercially necessary to generate the income from those services were all irrelevant in the absence of a direct and immediate link between the advertising and the relevant supplies. For example:

in Roald Dahl, the First-tier Tribunal held that, where a taxable person carried on two separate activities, one at a profit and one at a loss, and used the profits from the former to subsidise the losses on the latter, that did not mean that the costs of carrying on the former had a direct and immediate link with the latter. Similarly, just because one activity was being run at a loss for the purpose of generating customers for a profit-making activity did not mean that the costs of carrying on the former were costs of carrying on the latter; and

in ROH, the fact that the catering revenue was used to fund the production costs and that the production costs increased the sales of catering were not sufficient to create a direct and immediate link.

The tax treatment did not depend on the mere existence of substantial commercial links between the two activities;

this could be seen in ROH at paragraphs [38] and [83] to [89]. In ROH, the appellant sought to rely on the substantial economic links between the production costs and the supplies of catering to establish that there was a direct and immediate link between the two. However, in rejecting that submission, the Court of Appeal said that the references in Sveda and The Commissioners of Her Majesty’s Revenue and Customs v Associated Newspapers Limited [2017] EWCA Civ 54 (“ANL”) to an economic link “do no more than explain the nature of the connection required to satisfy the test of a direct and immediate link in cases where there is also a link with non-economic activities, such as the gratuitous provision of the path and of the retail vouchers. They do not herald a new and broader test for determining the existence of a direct and immediate link”. It rejected the appellant’s claim because it agreed with the Upper Tribunal in that case that, notwithstanding the substantial economic link between the production costs and the catering supplies, that link was not direct and immediate. The fact that the production costs enabled the appellant to make its catering supplies did not suffice and nor did the fact that the revenue from the catering was used to fund the production costs;

with regard to the point made by Ms Barnes and Mr Reynolds in paragraph 160(3)(a) above, the Court of Appeal’s reliance in DaP on the financial significance of the insurance income to the appellant’s business needed to be considered in the light of the fact that, in that case, the adverts had referred to the availability of insurance and had been intended to promote the insurance to customers and the appellant had sought to counter the importance of that fact by downplaying the importance to it in financial terms of the insurance relative to the supplies of mobile phones. Thus, the decision in DaP was not authority for the proposition that the importance of the revenue stream alone could serve to establish a direct and immediate link to supplies which were not even mentioned in the advert; and

the decision in N Brown should be approached with caution in the context of these appeals because:

it was a first instance decision which had not been tested on appeal;

it pre-dated the decision in ROH and so should be read in the light of the decision in that case;

there was a two-way relationship between the supplies of goods and the supplies of credit in that case which was not the same in this case. There was no suggestion on the facts of this case that the existence of the insurance offering was a factor which increased the sales of sofas; and

there was no authority in the case law for the suggestion in N Brown that advertising created a different test in this context from the test applicable to other costs such as production costs or the costs of purchasing goods.

The nature of the test – closely linked

162. Pursuing a similar theme, Ms Barnes and Mr Reynolds said that the two categories of supplies in this case were so closely intertwined that it made no sense to say that there was a direct and immediate link to one but not the other. For instance:

a customer buying a sofa was, subject to some minor exceptions, always offered insurance;

in respect of Sofology, customers had to opt out of insurance, and not opt in, at the point when the sofa was sold;

in respect of DFS, the staff were incentivised through commissions to offer insurance at the point when the sofa was sold;

the majority of customers who bought a sofa also bought insurance and, as the attachment rates were fairly constant in each case, that meant that the Appellant could predict that outcome with a reasonable degree of certainty;

the sale process was seamless – there was a single customer journey and information gleaned during the negotiations for the sofa was used by the sales staff in order to sell the insurance;

a reduction in sales of insurance would reduce the marketing budget and hence, if all other market factors remained constant, lead to a reduction in sales of sofas; and

in the case of Sofology, the same marketing strapline was used to sell both sofas and insurance.

163. In response, Ms Sloane said that the fact that there was a concerted programme of selling insurance to nearly everyone who purchased a sofa did not mean that, in addition to the direct and immediate link between the cost of the adverts and the supplies of sofas, there was a direct and immediate link between the cost of the adverts and the supplies of insurance intermediary services because that was no different from the fact pattern in ROH. In that case, the business used the production costs to sell tickets and catering. At the point of purchasing a ticket to the performance, a prospective customer was automatically offered champagne. A concerted add-on strategy did not create a direct and immediate link. Only the costs associated with the marketing and promotion of the add-on were directly and immediately linked to the supply of the add-on. This was not a case where the advert sold the insurance or the insurance somehow sold itself. Instead, the advert sold the sofa and the quality of the insurance product, the skill of the sales staff and the promotional videos and literature which related to the insurance sold the insurance. The high attachment rate was attributable to those matters and those matters alone. The high attachment rate said nothing about whether the cost of the adverts was directly linked to the supplies of insurance intermediary services. In DFS1, the attachment rate for fabric sofas was in fact very high – some 87% - and that had made no difference to the outcome of that case, which the Respondents accepted had been correctly decided.

164. Moreover, the attachment rates for each Appellant varied significantly as between, on the one hand, in-store sales and, on the other hand, online and telephone sales, and yet the Respondents had not accorded any relevance to this point. In fact, it demonstrated very well that it was not the cost of the adverts but rather the cost of the skilled sales staff which had the direct and immediate link to the supplies of insurance intermediary services. The fact that the overall attachment rate was predictable just meant that each Appellant had a commercial opportunity arising out of the sofa sales which it made and had a good idea of how valuable that commercial opportunity would be likely to be. No doubt the appellant in ROH had a good idea of the likely take-up of catering in relation to each production and that said nothing about whether or not there was a direct and immediate link between the production costs and the catering.

165. Thus, in this case, the costs which had a direct and immediate link with the supplies of the insurance intermediary services included things like the cost of making the video in relation to the insurance, the cost of the leaflets outlining the benefits of having insurance which were handed out in-store and the cost of training the sales staff better to sell the insurance. It did not include the cost of the PPC advertising which had nothing to do with those supplies. In fact, as was the case in Lok’nStore (see paragraph [57] in that case), the bulk of the costs which had the direct and immediate link with the supplies of insurance intermediary services in this case comprised the salaries of the sales staff (which did not include input tax) because it was the skill of the sales staff above all which directly led to the supplies in question.

166. The position in this respect was the same regardless of whether:

a skilled sales-person used information gleaned in the negotiations for the sale of one item to improve the prospects of selling another item. That was a fairly basic sales technique and said nothing about whether the costs which had a direct and immediate link to the supply of the first item were also directly linked to the supply of the second;

as was the case in relation to Sofology, the same strapline was used to sell the second item as was used to sell the first item. That again was simply a basic sales technique which shed no light on the answer to the question; and

the customer was given the opportunity to opt in to the second item, or, as was the case in relation to Sofology, the second item was automatically added to the order unless the customer opted out. The second item was an add-on in both cases, and the “opt-out” scenario was merely a harder selling technique than the “opt-in” scenario. From the commercial perspective, it made it more likely that the advert would lead to a sale of the second item but it did not create a direct and immediate link between the advert and that sale because the advert continued to have a direct and immediate link solely to the sale of the first item, which sale was a necessary pre-condition to that sale.

Summary in relation to the primary argument

167. In summary, whilst Ms Barnes and Mr Reynolds were of the view that the substantial economic links between the adverts and both types of supply, coupled with the content and layout of the landing pages, meant that the cost of the adverts had a direct and immediate link with both types of supply, Ms Sloane submitted that substantial economic links were not sufficient in and of themselves to establish a direct and immediate link between the cost of the adverts and the supplies by each Appellant of insurance intermediary services and that the other circumstances in this case – most notably, the content of the adverts, the content and layout of the landing pages and the manner in which the insurance was sold by each Appellant – showed that the links between the cost of the adverts and the supplies of insurance intermediary services was indirect.

The secondary argument

168. Ms Barnes and Mr Reynolds said that the Respondents’ alternative case was that there was no direct and immediate link between the cost of the adverts and any of the supplies made by each Appellant (whether of sofas or anything else) for the following reasons:

in Mayflower, Carnwath LJ had made it clear that:

whether or not costs were overheads for VAT purposes was a binary choice and not a spectrum; and

the overheads category served a particular, albeit limited, purpose in the VAT system; and

in this case, as noted in paragraph 160(9) above, the nature of the PPC advertising was such that it was all about getting users to the relevant Appellant’s website and promoting the relevant Appellant’s brand as a whole. The home page, which was most often the landing page, was very general in nature and the relevant Appellant clearly did not intend the user to stay on that page. In addition, there were many different paths that a user might take upon leaving that page. These factors suggested that it was not possible to see the cost of the adverts as having a direct and immediate link with any particular category of supply.

169. Ms Sloane said that the cost of the adverts in this case was not an overhead. There was a very obvious direct and immediate link between the cost of the adverts and the supplies of sofas. That link was identifiable by reference to the same criteria as had been adopted in Skipton and DaP. In other words, objectively, the adverts – whether they were “shopping” or “search” – related very clearly to sofas. There was no need for the link in question to be with a specific model of sofa, just as an advert for mobile phones did not need to be linked with a specific model of mobile phone in order for there to be a direct and immediate link with the supply of mobile phones. The mere fact that a cost related to a category of items instead of a specific item did not make that cost an overhead. Moreover, that link existed even if the intention (and effect) of the advert was to entice the user onto the website or into a store in order to purchase the relevant category of items. That was true of all advertising. If the intention and effect was to entice the user onto the website or into the store for the purpose of buying the particular category of goods, then that direct and immediate link existed. It would of course be different if the intention and effect was simply to entice the user onto the website or into the store for some general purpose which was unrelated to the sale of a particular category of goods. The latter was very much not the case on the present facts.

170. Since there was a direct and immediate link between the cost of the adverts and the supplies of sofas, the cost of the adverts was not an overhead of either Appellant’s business.

discussion

Conclusions

171. Before setting out our conclusions in relation to the two issues which we have been asked to address, we would like to acknowledge the clarity and forensic skill demonstrated by each of the three counsel for the parties in this case, to which we pay tribute. We are indebted to them for the cogency with which they have drawn out the issues and the various arguments that we have needed to consider but, having reflected on their respective submissions and those issues, we think that the answers to the questions posed to us are clear.

172. Despite the persuasive arguments made to the contrary by Ms Barnes and Mr Reynolds, our conclusions are that, for the reasons which are set out in the paragraphs which follow:

for each Appellant, the cost of the PPC advertising had a direct and immediate link with the taxable supplies of sofas made by the relevant Appellant and did not have a direct and immediate link with the exempt supplies of insurance intermediary services made by the relevant Appellant;

in consequence of the conclusion set out in paragraph 172(1) above, for each Appellant, the cost of the PPC advertising was not an overhead of the relevant Appellant’s business or an overhead of a discrete part of the relevant Appellant’s business; and

as a result, each Appellant’s appeal should be upheld.

173. The above conclusions are based on the analysis which follows.

Direct and immediate link

174. The starting point in our analysis is to set out the principles which need to be taken into account in identifying whether a particular cost has a direct and immediate link with a particular supply. The phrase “direct and immediate link” is deceptively simple, as the proliferation of case law in this area has demonstrated. The difficulty is that the application of the phrase is acutely fact-sensitive, so that even apparently small differences in fact patterns can give rise to very different results.

175. The relevant principles may briefly be summarised as follows:

the answer to the question of whether or not a direct and immediate link exists between a cost and a supply “is to be objectively ascertained from the facts and circumstances of the transactions, not by investigating the subjective intentions of the taxable person” – see ROH at paragraph [17]. This means that, although the test is multi-factorial in nature, and it is appropriate to take into account, as one of those factors, the purpose of the taxable person, that purpose is to be identified objectively from the facts and circumstances and not by reference to the subjective intentions of the taxable person;

similarly, in determining whether a cost has a direct and immediate link with a supply, it is not appropriate to look at the ultimate aim of the taxable person when incurring the relevant cost or whether or not the cost is reflected in the price charged by the taxable person for the relevant supply – see BLP at paragraphs [19] to [21] and [24] and DaP at paragraphs [20] to [24];

in addition, the fact that a supply would not have been made by the taxable person but for a cost incurred by the taxable person is not sufficient to create a direct and immediate link between the cost and the supply. The relevant test is not a “but for” test – see Southern at paragraphs [32] to [37] and DaP at paragraphs [34] to [36];

the fact that there is a “close economic link” or a “necessary economic link” between a cost incurred by a taxable person and a supply made by the taxable person is not sufficient to create a direct and immediate link between the cost and the supply – see ROH at paragraphs [80] et seq.. Although there are decisions such as Sveda and ANL where the phrases “close economic link” or “necessary economic link” have been used in the course of describing a direct and immediate link between a cost incurred by the taxable person and a supply made by the taxable person, those decisions were dealing with circumstances where there was a more immediate non-economic activity between the incurring of the cost by the taxable person and the supply made by the taxable person - the gratuitous provision of a path in Sveda and the provision of free vouchers in ANL – and the relevant court used those phrases to explain why the intermediate non-economic activity did not prevent the direct and immediate link between the cost and the supplies from arising. Those cases are therefore highly fact-specific and “do not herald a new and broader test for determining the existence of a direct and immediate link” – see ROH at paragraphs [81] to [83];

the above means that, even if a cost which is incurred by the taxable person is essential in economic terms to a supply made by the taxable person, there may still not be a direct and immediate link between the two – see ROH at paragraphs [84] to [88]. For example, whilst, in Mayflower, there was a direct and immediate link between the cost of buying in productions and the supplies of programmes (because the former provided the content for the latter), no such direct and immediate link existed in ROH between, on the one hand, the cost of the productions, and, on the other hand, the catering services despite the economically-interconnected nature of those supplies. This was because the production costs were not used to make the supplies of catering but were instead used solely for putting on the productions – see the decision of the Upper Tribunal in The Commissioners of Her Majesty’s Revenue and Customs v Royal Opera House Covent Garden Foundation [2020] STC 1170 (“ROH UT”) at paragraphs [106] to [109], cited with approval by the Court of Appeal in ROH at paragraphs [33] and [88];

as long as there is a direct and immediate link between a cost incurred by the taxable person and an exempt supply by the taxable person, the right to deduct input tax on that cost will be restricted and that will be the case even if, in addition to that direct and immediate link, there is also a direct and immediate link between that cost and a taxable supply by the taxable person and the direct and immediate link between the cost and the taxable supply is more direct and immediate than the direct and immediate link between that cost and the exempt supply – see the opinion of Advocate General Jacob in Abbey National at paragraph [35] and DaP at paragraph [30]; and

finally, although it is of no relevance on the basis of the facts in the present appeal, there are circumstances in which a direct and immediate link between a cost and a supply made as part of a chain of supplies can be broken by an exempt supply made earlier in that chain – see Sveda at paragraphs [32] to [34], RAC at paragraph [43] and ROH at paragraph [91].

176. The above are the relevant principles of law and it is then necessary to apply those principles to the facts in the present case.

Physical versus economic links

177. Before addressing specifically the two arguments made by the Respondents in this case, there are two preliminary points which we should make.

178. The first is as follows. At the hearing, much was made by the parties of the dichotomy between physical links and economic links when it came to identifying the supplies with which the PPC advertising costs had a direct and immediate link. For instance, Ms Barnes and Mr Reynolds described Ms Sloane’s approach of focusing primarily on the content of the adverts and the content and layout of the landing pages as amounting to a focus on physical links to the exclusion of economic links. They said that this approach was overly-reductive and that the economic links between the adverts and the supplies of insurance intermediary services were a meaningful factor in the complete circumstances which were required to be taken into account.

179. We do not find this dichotomy between physical links and economic links to be particularly helpful in dealing with the questions which we have to address. That is because:

describing the content of the adverts and the content and layout of the landing pages as “physical” seems to us to disregard the virtual nature of those items;

more importantly, as the substance of the arguments before us made clear, it is impossible to address the question of whether or not the cost of an advert has a direct and immediate link with a particular supply without considering the economic consequences of the advert. After all, if an advert has increased the quantum of the supplies of a particular item, that economic consequence is undoubtedly an important part of the circumstances in which any objective view of whether or not there is a direct and immediate link between the advert and the supplies is to be reached. As the First-tier Tribunal noted in N Brown, marketing material is used to stimulate demand for one or more of the goods and/or services which the retailer is offering and it is therefore essential to identify the economic consequences of that material if one is to identify the supplies with which the cost of the material has a direct and immediate link;

given the above, we consider that, in each case, the relevant question is not whether the link between the relevant cost and the relevant supply is physical or economic but rather whether the link between the relevant cost and the relevant supply, whatever form it takes, is direct and immediate or is instead indirect. In other words, in the case of an economic link between a cost and a supply, that economic link could be direct and immediate or could instead be indirect. In both cases, the link in question would be an economic link but only in the former case would the input tax on the cost be attributed to the relevant supply; and

in this context, we think that the question of whether an economic link is direct and immediate or is instead indirect does not turn on the strength or extent of the economic link. It turns instead on the precise nature of the economic link. A very strong and substantial economic link could still be indirect, just as a very weak and insubstantial economic link could be direct and immediate. It all depends on the precise nature of the link. For example, in ROH, there was a strong and substantial economic link between the production costs and the catering supplies. Indeed, the economic link was described as being “essential” (see ROH UT at paragraph [98]). And yet both the Upper Tribunal and the Court of Appeal concluded that the link was not direct and immediate and therefore could not result in an attribution of the input tax on the production costs to the catering supplies.

180. The second point arises out of the first and is as follows. For the reasons set out in paragraph 179 above, we think that the debate between the parties as to whether or not those decisions in which the relevant court took into account economic links at the apportionment stage of the process of determining input tax recovery apply to the attribution stage in the process as well as to the apportionment stage did not take matters very far forward. As will be apparent from the view set out in paragraph 179 above, we consider that economic links are factors which should be taken into account at both stages in the process of determining input tax recovery and not just at the apportionment stage. The economic link between a cost and a supply is as much part of the full facts and circumstances which need to be considered in determining whether the cost can be attributed to the supply as any other facts and circumstances which might shed light on the question of whether the link between the two is direct and immediate. Thus, it needs to be taken into account at the attribution stage in the process. However, the analysis of the economic link between the cost and the supply at the attribution stage should be focused on whether the link is direct and immediate or is instead indirect. It should not be focused on the strength or extent of the link. That is because the former is the only question which is relevant at the attribution stage. Slightly different considerations should predominate at the second stage of the process, when the strength and extent of the economic links to each category of relevant supply become much more relevant.

The secondary argument – overheads

181. With these general points in mind, we now turn to address the two specific arguments which were made by the Respondents in this case.

182. In this regard, although it was only the fallback argument made by the Respondents, we think that the most logical starting point is to consider whether the PPC advertising costs incurred by each Appellant in this case amounted to overheads. We say that because:

those costs could not have amounted to overheads if they had a direct and immediate link with the relevant Appellant’s supplies of sofas (regardless of whether or not they also had a direct and immediate link with the relevant Appellant’s supplies of insurance intermediary services); and

we think that it is clear that those costs did have a direct and immediate link with each Appellant’s supplies of sofas.

183. In addressing this question, we are mindful of the admonition by Carnwath LJ in Mayflower at paragraph [33] to the effect that the overheads category “serves a particular and limited purpose in the VAT system, for those inputs which would not otherwise be brought within the calculation. It should not be extended beyond that purpose”. It follows that the allocation of costs to the category of overheads is a last resort to be adopted only if it is not possible to identify a supply or supplies with which the relevant costs have a direct and immediate link.

184. We have previously found as facts that:

the purpose of each Appellant in purchasing PPC advertising was to encourage users:

to enter its website, as opposed to the websites of its competitors; and

to go into its stores, as opposed to the stores of its competitors,

in each case, for the purpose of acquiring sofas; and

we reached this conclusion because:

that was each Appellant’s marketing strategy in purchasing PPC advertising. The strategy was directed at increasing sales of sofas and building awareness of the relevant Appellant’s sofa offering;

each of the “shopping” adverts contained photographs of sofas and each of the “search” adverts referred to sofas as part of the text; and

when clicking on an advert, the user was taken to a landing page which showed images of sofas.

185. On the basis of those facts, we have concluded that, viewed objectively, there was a direct and immediate link between the PPC advertising costs incurred by each Appellant and the supplies of sofas made by the relevant Appellant, with the result that the PPC advertising costs were not part of the relevant Appellant’s overheads. Even if one were to exclude each Appellant’s marketing strategy from the relevant factual matrix to be taken into account in determining this question, on the basis that that was no more than the relevant Appellant’s subjective intention, the other facts set out in paragraph 184(2) above are sufficient in our view to establish objectively the existence of that direct and immediate link. (We would add that, in any event, we think that each Appellant’s marketing strategy is also a fact which should properly be taken into account in determining this question as it extended beyond the subjective purpose of any one or more individuals at each Appellant and was instead ingrained within the business policies and strategies of the relevant Appellant. As such, it is in our view an objective fact which is entirely appropriate to be taken into account in reaching the objective determination in relation to each Appellant and, when one does so, one sees that it serves only to reinforce the conclusion which we have reached on the basis of the other facts set out in paragraph 184(2) above.) In relation to this conclusion, we agree with Ms Sloane that:

in determining whether there is a direct and immediate link between the PPC advertising costs and the supplies of sofas made by the relevant Appellant, it is not necessary to establish that there was a direct and immediate link between the costs and the supply of a specific model of sofa. It is instead sufficient for this purpose to establish that there was a direct and immediate link between the costs and the supply of a sofa of some sort. A promotional link between the advert and the Appellant’s sofas in general is perfectly sufficient; and

it is irrelevant that the purpose of each Appellant in incurring the PPC advertising costs was to entice the user onto the website or into a store. This is because it was a critical part of that purpose that the user would go onto the website or into a store specifically in order to purchase a sofa. It might have been different if the purpose of the relevant Appellant in incurring the costs had simply been to entice the user onto the website or into a store for some general reason other than the specific reason of purchasing a sofa.

186. The Respondents’ submission to the contrary in relation to this secondary argument was that the nature of the PPC advertising was such that it was all about promoting the relevant Appellant’s brand as a whole and enticing users to the relevant Appellant’s website for the purpose of maximising the relevant Appellant’s sales in general. Consistent with that submission, Ms Barnes and Mr Reynolds pointed out that:

the nature of PPC advertising was such that the content of the adverts had a minimal role to play in determining whether there was a direct and immediate link between the cost of the advert and any particular category of supply by the relevant Appellant. Instead, the content and layout of the landing pages was of greater importance;

in most cases, the landing page offered the user access to a wide variety of different products which were being offered by the relevant Appellant and was intended merely as a starting point in a customer journey which could lead to the purchase of a number of different items. For example, the home page, which was most often the landing page, was very general in nature, the relevant Appellant clearly did not intend the visitor to stay on that page and there were many different paths that a visitor might take upon leaving that page;

looking at the economic position, the purpose of each Appellant in incurring the PPC advertising costs was to maximise the sales of all of its products and not one specific type of product. To that end, each Appellant was intent on maximising its revenues, profitability and average order values and, in that respect, it measured its success or failure by aggregating the revenue streams from a variety of products and did not distinguish between the different categories of product; and

the above factors suggested that it was not possible to see the adverts as having a direct and immediate link with any particular category of supply.

187. In relation to the points made in paragraph 186 above:

we do not agree that the nature of PPC advertising means that little weight can be accorded to the terms of the advert. In our view, the fact that the relevant Appellant did not have to pay for the service unless and until the user clicked on the advert is neither here nor there. The nature of the service for which the relevant Appellant was paying Google was the impression of its advert on the Google platform. It was the content of that impression which induced the user to click on the link in the first place. We therefore consider that, whilst the content of the adverts was not solely determinative of whether or not there was a direct and immediate link with any particular category of supply in this case, that content was a highly relevant factor, as in fact it was in DaP. It is impossible properly to determine objectively whether there was a direct and immediate link between the cost of the adverts and any particular category of supply without taking into account the content of the adverts;

whilst we agree with Ms Barnes and Mr Reynolds that the content and layout of the landing pages were also relevant factors to be taken into account in determining whether or not there was a direct and immediate link between the PPC advertising costs and any particular category of supply in this case, we do not agree with their characterisation of the landing pages set out in paragraph 186(2) above. As the findings of fact referred to in paragraph 184 above make clear, we think that the purpose of the relevant Appellant in incurring the costs was to entice the user to visit the website or store of the relevant Appellant specifically in order to purchase a sofa. That was why the adverts either pictured or referred to sofas and why the landing pages contained images of sofas. It was not the case that the relevant Appellant merely wanted to entice the user to visit the website or store for some general purpose – which is to say, to purchase any of the items which the relevant Appellant happened to be offering for sale at the time when the user entered the website. The focus of the PPC advertising costs was very much on the supplies of sofas; and

we have therefore concluded that, viewed objectively, the PPC advertising costs had a direct and immediate link with each Appellant’s supplies of sofas. In that regard, we do not think that the economic position referred to in paragraph 186(3) above affects the conclusion in any way. Naturally, the relevant Appellant would have been delighted if the user, having entered the website or store to purchase a sofa, were to have purchased items in addition to the sofa. And it is not surprising that the relevant Appellant took steps to increase the likelihood that that would be the case. After all, each Appellant was in business to maximise its revenues, profitability and average order values and any items which it could sell in addition to the sofa would have had a beneficial impact on those figures. However, viewed objectively in the light of all the facts, that was not the reason why the relevant Appellant incurred the cost of the adverts. Each Appellant incurred the cost of the adverts specifically in order to promote its sales of sofas.

188. The fact that there was a direct and immediate link between the cost of the adverts and the supplies of sofas made by the relevant Appellant means that this secondary argument on the part of the Respondents fails.

The primary argument – direct and immediate link with the supplies of insurance intermediary services

189. The Respondents’ primary argument is that, even if the PPC advertising costs did have a direct and immediate link with the supplies of sofas which were made by each Appellant, they also had a direct and immediate link with the supplies of insurance intermediary services which were made by each Appellant.

190. In relation to that argument, we would agree with the submissions made by Ms Barnes and Mr Reynolds as to the importance to each Appellant, from the commercial perspective, of the supplies of insurance intermediary services. No-one could doubt the economic significance to each Appellant of those supplies. Both the margin and the revenue generated by the services was substantial. For that reason, each Appellant took significant steps to monitor and to maximise the attachment rates. The evidence which was presented to us made that very clear and we have made findings of fact to that effect in paragraphs 142(6) and 142(7) above.

191. Similarly, we would agree with Ms Barnes and Mr Reynolds that, for each Appellant, there was a substantial economic connection between the adverts and the supplies of insurance intermediary services. For each Appellant, as the attachment rates were relatively constant, an increase in sofa sales was likely to translate into an increase in insurance intermediary commissions.

192. We also agree that, for each Appellant, there was a close link between the supplies of sofas and the supplies of insurance intermediary services. For instance:

a customer purchasing a sofa would, subject to some minor exceptions, always be offered insurance;

in respect of Sofology, customers had to opt out of insurance, and not opt in, at the point when the sofa was sold;

in respect of DFS, the staff were incentivised through commissions to offer insurance at the point when the sofa was sold;

the sale process was seamless – there was a single customer journey and information gleaned during the negotiations for the sofa was used by the sales staff in order to sell the insurance;

a reduction in sales of insurance would reduce the marketing budget and hence, if all other market factors remained constant, lead to a reduction in sales of sofas; and

in the case of Sofology, the same marketing strapline was used to sell both sofas and insurance.

193. However, as Ms Sloane concisely put it, commercial and economic links, no matter how substantial, do not convert an indirect link into a direct and immediate link. The decision in ROH shows that a substantial economic link is not sufficient, in and of itself, to amount to a direct and immediate link. The question which needs to be addressed is whether that substantial economic link amounts to a direct and immediate link and is not simply an indirect link.

194. We have concluded that, when one takes into account all the facts and circumstances of this case, and views the position objectively, it is apparent that, for each Appellant, the link between the advertising costs and the supplies of insurance intermediary services, no matter how substantial it was in economic terms, was indirect in nature. We say this for the following reasons:

first and foremost, when viewed objectively, the adverts were designed to attract users to enter the relevant Appellant’s website and/or to go into the relevant Appellant’s stores specifically for the purpose of buying sofas. Even if one were to ignore each Appellant’s marketing strategy on the ground that it was no more than subjective intention – an approach which we do not condone, for the reason set out in paragraph 185 above - that much is apparent from the content of the adverts and the content and layout of the landing pages. Each of the “shopping” adverts contained photographs of sofas and each of the “search” adverts referred to sofas as part of the text and, when clicking on a “shopping” or “search” advert, the user was taken to a landing page which showed images of sofas.

Conversely, neither Appellant incurred the cost of the adverts with the purpose of promoting its insurance offering. Again, even if one were to ignore the fact that each Appellant’s marketing strategy was not directed specifically at increasing sales of insurance, the other facts in this case point firmly in that direction. For example, the fact that the relevant Appellant offered insurance was not a feature which it sought to emphasise in its adverts, the relevant Appellant did not bid in the Google auction for words relating to insurance and insurance did not feature as a search term in any of the relevant Appellant’s adverts. In addition, a user clicking on a “shopping” or “search” advert was never taken to a landing page that highlighted the relevant Appellant’s insurance offering. Although it was possible to navigate from the landing page to the page on the relevant Appellant’s website which dealt with insurance, the link to get to the insurance page was not a prominent feature on the landing page. The link had to be located by hovering over the “More” or “Help” item on the navigational toolbar and then scrolling down and clicking.

Moreover, unless the user chose to navigate to the page which dealt with the relevant Appellant’s insurance offering, insurance was not part of the user’s journey on the website until after the sofa had been added to the user’s basket.

These facts are some distance away from the facts in DaP, where the availability of free insurance was expressly highlighted in the advert and was held to have been intended to attract new customers to the relevant insurer.

In relation to this point, we think that it is of no relevance that a user who was taken to the Sofology home page might, in some instances, be required to make two clicks to get to an individual product page but only one click to get to the Sofashield page – see paragraph 82 above. In our view, the precise number of clicks which might have been required to get to a particular page was of no moment in relation to this question, in the same way that, in the context of a customer visiting a store as a result of an advert, the mere fact that the advertised product was on the second floor and another product happened to be on the first floor would be of no moment. The key fact is that Sofashield did not feature prominently on the Sofology home page and had to be sought out by the user, if desired;

secondly, we think that a highly significant fact is that it was not possible for a user to purchase insurance in isolation. Insurance could be purchased only in conjunction with the purchase of a sofa, and consequent upon the user’s decision to purchase the sofa. The link between, on the one hand, the cost of an advert which was intended to promote the supply of a sofa and, on the other hand, a supply of insurance intermediary services could therefore only be indirect. The supply of insurance intermediary services was necessarily indirect because the supply of the sofa was a necessary pre-condition to its existence. That accounts for the features of the adverts, the landing pages and the website described in paragraph 194(1) above. In short, when viewed objectively, the aim of the PPC advertising was to induce users to purchase sofas. Only after the user had decided to purchase a sofa could the user then be sold insurance. The link to the supply of the sofa was direct and immediate. The link to the supply of the insurance intermediary services was indirect.

In this regard, the fact that each Appellant was aware that the attachment rates remained fairly constant over time so that an increase in sofa sales would be likely to lead to an increase in the commissions received by the relevant Appellant for its insurance intermediary services is, in our view, irrelevant, in the same way that, in Roald Dahl, it would have been irrelevant if the museum shop had been able to increase the sales of certain items which it sold because the exhibition attracted visitors. The relative constancy of the attachment rates merely meant that there was an economic correlation between revenues arising from sales of sofas and the insurance intermediary commissions. It did not create a direct and immediate link between the cost of adverts whose purpose was to increase sales of sofas and the supplies of insurance intermediary services which were related to, and dependent upon, the sales of sofas;

thirdly, the indirect nature of the link between the advertising costs and the supplies of insurance intermediary services was apparent in the way in which each Appellant went about selling the insurance. For instance, sales staff were trained that, unless the subject of insurance was raised by the prospective customer himself or herself at an earlier stage, insurance was not to be raised with the prospective customer until he or she had decided to purchase a sofa, for fear of losing the sofa sale; and

finally, far from being an argument in favour of establishing that the link between the cost of the adverts and the supplies of insurance intermediary services was direct and immediate, as the Respondents sought to use them, we think that the steps which each Appellant took to promote its insurance offering by means other than the adverts themselves demonstrates that the link between the cost of the adverts and the supplies of insurance intermediary services was in fact indirect. We have referred in paragraph 142(6) above to the various ways in which each Appellant sought to maximise the revenues it derived from its supplies of insurance intermediary services. Those steps were that:

for each Appellant, sales staff were trained to use information provided by the customer during the negotiations in relation to the sale of the sofa to push the insurance and to ensure, so far as allowed by the prevailing regulatory regime from time to time, that customers were encouraged to add the insurance to their sofa purchase;

for each Appellant, the attachment rates were high and the attachment rate for in-store sales was higher than the attachment rates for online or telephone sales, thereby demonstrating the impact on attachment rates made by the skill of the sales staff and the power of face-to-face personal persuasion on the part of the sales staff;

each Appellant’s marketing strategy involved driving footfall into stores, where the attachment rates were higher;

each Appellant monitored attachment rates as one of its KPIs;

in the case of Sofology, the company operated an “opt-out” basis in relation to sales of insurance; staff were instructed to include the insurance on the invoice and wait for the prospective customer to opt out;

in the case of Sofology, the company took steps to ensure that there was a close connection between its brand and its insurance offering;

in the case of DFS, staff were incentivised to sell insurance by being paid commissions which took the sales of insurance into account; and

each Appellant produced leaflets and/or videos extolling the benefits of the insurance.

In our view, these facts demonstrate that what directly generated the sales of insurance was what happened after the prospective customer had resolved to purchase a sofa in consequence of clicking on an advert. Putting this another way, we agree with Ms Sloane that the high attachment rate was attributable to the quality of the insurance product, the skill and training of the sales staff and the promotional videos and leaflets which related to the insurance. This was well-demonstrated in the fact that there tended to be a dip in the in-store attachment rates at DFS in the run up to Christmas when the in-store sales staff were stretched and additional assistants who were not trained to sell Sofacare had to be drafted in.

The reasoning set out above means that, in our view, the costs which had a direct and immediate link with the supplies of the insurance intermediary services included things like the cost of making the videos in relation to the insurance, the cost of the leaflets outlining the benefits of having insurance which were handed out in-store, the cost of the training given to staff on how to promote the insurance and, of course, the staff salaries. In contrast, the advertising costs had only an indirect link with the supplies of the insurance intermediary services because no sale of insurance could occur until the intermediate step of persuading the prospective customer to purchase a sofa had occurred and it was these other steps which then operated to generate the sale of insurance.

The position in this case is not dissimilar in our view from the postcards featuring photographs of museum exhibits which were sold in the museum shop in Roald Dahl. In that case, there was a direct and immediate link between the exhibition costs and the exhibits. The photographs of the exhibits which were then put onto the postcards were an additional step which made any link between the exhibition costs and the postcards indirect.

In this regard, the matters referred to above in paragraphs (a) to (h) of this paragraph 194(4) do no more than reveal that, from the commercial perspective, each Appellant recognised that there was considerable value to be derived from selling insurance to as many customers as possible who bought sofas. They say nothing about whether PPC advertising costs which had a direct and immediate link with the supplies of sofas also had a direct and immediate link with the supplies of insurance intermediary services. From the commercial perspective, those matters made it more likely that the PPC advertising costs would lead to the supplies of insurance intermediary services but that was because of the direct and immediate impact which those matters had on such supplies. That contrasts with the indirect impact which the PPC advertising costs themselves had on the supplies of insurance intermediary services.

195. We should stress that we do not think that this is an example of a case where a cost has a direct and immediate link with two distinct categories of supply but has a closer direct and immediate link with one category of supply than the other. We recognise that there may be circumstances in which that is capable of being the case, as mentioned by Jonathan Parker LJ in DaP at paragraph [30]. The existence of a closer direct and immediate link between a cost and a particular category of supply does not preclude the possibility of a more distant, albeit sufficient, direct and immediate link between the relevant cost and another category of supply. However, we consider that this is not an example of such a case. Here, we are not saying merely that the link between the PPC advertising costs and the supplies of sofas was closer than the link between the PPC advertising costs and the supplies of insurance intermediary services. Instead, we are saying that the link between the PPC advertising costs and the latter was indirect because:

the latter was not a driver for the adverts – it was not the reason for the adverts, it did not feature in the search terms, it was not highlighted on any landing page and it was not the reason why users clicked on the adverts;

the latter could not take place unless and until the former took place and staff were instructed not to raise the subject of insurance until the former was assured; and

there were other matters - such as the quality of the insurance product, the marketing material relating to the insurance product and the training given to, and actions taken by, the staff – which had a direct and immediate link with the latter and whose relationship with and impact on the latter serve to emphasise the indirectness of the link between the PPC advertising costs and the latter.

196. Turning then to deal with the various points made by Ms Barnes and Mr Reynolds in opposition to the above conclusion, we would comment as follows:

as we have already noted in paragraph 175(1) above, we do not agree that the taxable person’s subjective intentions form part of the factual matrix which is to be taken into account in determining whether a cost has a direct and immediate link with a particular supply. To be clear, we consider that, in each case, the taxable person’s purpose most certainly is part of the relevant factual matrix. However, as noted by David Richards LJ in the passage from ROH cited in paragraph 175(1) above, that purpose is to be ascertained objectively from the facts and circumstances of the transactions and not by investigating the subjective intentions of the taxable person. Only by adopting that objective approach can one escape the error of taking into account subjective intentions (see Warren J in St Helen’s at paragraph [77]). Moreover, it is important to bear in mind that, in examining objectively the taxable person’s purpose, the relevant question to be answered is whether that purpose reveals that there is a direct and immediate link between the relevant cost and a particular supply. It is not whether that purpose reveals that the relevant cost was intended to maximise the revenues arising from a particular supply. In that regard, we agree with Ms Sloane that, by its very nature, every business is intent on maximising the revenues arising from its expenditure and so the fact that the revenues arising from a particular supply may be a KPI for the business or be of great interest to the managers of the business does not say anything about whether the link between a particular cost and that supply is direct and immediate or is instead indirect;

a fundamental plank of the Respondents’ case was that a distinction could be made between cases like Mayflower and ROH – which pertained to production costs – and cases like DaP and the present case – which pertained to advertising costs.

Ms Barnes and Mr Reynolds submitted that, in the production costs cases, the focus was necessarily on the category of supplies to which the production costs directly gave rise. For example, in ROH, since it could not be said that the production costs directly gave rise to the supplies of catering, the link between the two was insufficient to give rise to an attribution. In that sense, it could clearly be seen that the production costs were “promotional” of the ticket sales but were not “promotional” of the catering supplies. Instead, what led to the opportunity to make the catering supplies was not the production costs but the ticket sales. It was the ticket sales which brought the customers to the performances and thus it was they and not the production costs which were “promotional” of the catering supplies. As such, the extent of the economic ties between the production costs and the catering supplies could be disregarded, which is how the Court of Appeal approached the case.

In contrast, in the case of advertising costs, the extent of the economic ties between the relevant adverts and the supplies was highly relevant because the strength of the economic ties between the relevant adverts and the supplies was the most appropriate way of determining whether the relevant adverts were “promotional” of the supplies. This dichotomy in the way in which the test applied to different types of costs had been identified in N Brown, where the First-tier Tribunal contrasted the analysis applicable on a straightforward transaction of purchase and sale with the more wide-ranging economic analysis required in the case of advertising costs (see paragraph 47(5) above).

We do not agree that the test for identifying whether there is a direct and immediate link between a particular cost and a particular supply should be different depending on the nature of the cost. There is no authority binding on us to that effect and, to the extent that prior first instance decisions have adopted that approach, we do not agree with them. The test itself is always the same – which is to identify whether there is a direct and immediate link between a cost and a supply. The nature of the cost and the nature of the supply will undoubtedly affect the answer to that question but there is no authority for the proposition that somehow a more wide-ranging economic enquiry is appropriate in relation to advertising costs than is the case in relation to other costs such as production costs;

for the reasons which we have already set out in paragraph 187(1) above, we do not agree that the nature of PPC advertising means that little weight can be accorded to the terms of the adverts in this case. In our view, the fact that the relevant Appellant did not have to pay for the service unless and until the user clicked on the advert is neither here nor there. The nature of the service for which the relevant Appellant was paying Google was the impression of its advert on the Google platform and therefore the content of that impression, albeit not determinative, is a highly relevant factor in determining objectively whether the link between the cost of the adverts and the supplies of insurance intermediary services was direct and immediate. The position is no different from that pertaining in DaP;

having said that, we agree with Ms Barnes and Mr Reynolds that the content and layout of the landing pages was also a highly relevant factor to be taken into account in the objective determination of whether the link between the cost of the adverts and the supplies of insurance intermediary services was direct and immediate or was instead indirect. Indeed, we did not understand Ms Sloane to be disagreeing with that approach. After all, those pages were the first point of contact between the relevant Appellant and the user when the latter clicked on the advert. They were therefore bound to be highly relevant to the analysis.

However, as we have already noted in paragraph 194(1) above, we disagree with Ms Barnes’s and Mr Reynolds’s characterisation of the purpose and impact of the content and layout of the landing pages in this case. In our view, the landing pages operated in exactly the same way as the entrance to a store. In other words, the fact that it was possible to navigate away from a landing page to any of the relevant Appellant’s products was no different from the ability of a visitor to a store to ignore the first display to which he or she came and to explore the contents of the store more fully.

If insurance had featured prominently on the landing pages as a product – in other words, to pursue the store analogy, if insurance had been heavily promoted at the entrance to the store as the main item on entering the store – then that might well have affected the analysis. In that case, the example raised by the First-tier Tribunal in DFS1 of using a sprat to catch a mackerel might well have been relevant and that might well have meant that the content of the adverts was overwhelmed by the content and nature of the landing pages and thus of less significance. But, on the facts of this case, the mere fact that it was possible to navigate away from the landing pages to the insurance pages seems to us to be irrelevant.

The key facts are that the user clicked on an advert which was referable to sofas and was taken to a page on the website which had been selected by the relevant Appellant as being the most effective for selling sofas and which had images of sofas. It did not matter whether the sofa which the advert was designed to entice the user to buy was a specific sofa, a sofa falling within a specific broad category or a sofa of any kind. In contrast, insurance did not feature in the terms of the adverts and was simply one item amongst many which might be found on the landing pages, where it was given no prominence. The facts show plainly that the adverts, when seen in the light of the content and layout of the landing pages and the other facts mentioned in paragraph 194 above, had a direct and immediate link solely with the taxable supplies of sofas and did not have such a link with the exempt supplies of insurance intermediary services;

we do not agree that relying heavily on the content of the adverts and the content and layout of the landing pages in order objectively to identify the supplies to which the advertising costs had a direct and immediate link leads to absurdity or is overly-reductionist. On the contrary, we think that there is compelling logic in focusing on those facts as a means of identifying objectively the supplies with which the adverts had a direct and immediate link. Indeed, we think that the Respondents’ approach involves giving insufficient weight to the obvious pointers in that regard which are provided by the content of the adverts and the content and layout of the landing pages. Moreover, we should reiterate that we have not relied solely on those matters in reaching our conclusions. We have instead taken into account all of the relevant facts and circumstances, as outlined in paragraph 194 above;

contrary to the submission made by Ms Barnes and Mr Reynolds in paragraph 160(3) above, we consider that the Court of Appeal’s decision in DaP is not authority for the proposition that the financial significance of the income deriving from supplies to which costs have a link can suffice to make that link direct and immediate or is even relevant in determining whether that link is direct and immediate or is instead indirect.

The ratio of the Court of Appeal decision in DaP was contained in paragraph [73] of the decision. In that paragraph, Jonathan Parker LJ cited, with approval, the reasoning adopted by the VAT Tribunal in reaching its conclusion in paragraphs [49] and [51] of its decision (which he had previously set out in paragraph [43] of the Court of Appeal’s decision) to the effect that the adverts in that case, containing as they did references to the free insurance, were being used to attract customers not just to the airtime service contracts but also to the insurance. That ratio had nothing to do with the significance of the insurance intermediary income to the appellant. It was solely concerned with the fact that the adverts were being used to attract customers to the insurance, which is in our view a material distinction between the facts in DaP and the facts in this case.

Jonathan Parker LJ turned to address the significance of the insurance intermediary income to the appellant only in paragraphs [74] to [77] of the decision in the context of rejecting the submission which had been made by the appellant to the effect that the supplies of insurance intermediary services were, in a commercial sense, secondary to the supplies of airtime service contracts. In those paragraphs, Jonathan Parker LJ said, first, that whether or not they were secondary was irrelevant (see paragraph [74]) and, secondly, that they were in any event not secondary (see paragraphs [76] and [77]). At no point did he say that the significance of the insurance intermediary income would have been sufficient to found a direct and immediate link between the adverts and the supplies of insurance intermediary services in the absence of the promotional link between the adverts and those services which could be identified by the terms of the adverts.

In that regard, the fact that the adverts in DaP referred only to the three-month period of free insurance – in relation to which no supplies were made by the appellant – and not to the ongoing insurance after that period for which payments would need to be made and which would generate insurance intermediary commissions for the appellant is irrelevant. The key factual finding by the First-tier Tribunal was that the reference in the adverts to the three-month period of free insurance was clearly intended to attract new customers to the insurer so that the appellant would be able to earn its insurance intermediary commissions once the three-month free period was over. It was that which gave rise to the direct and immediate link between the adverts and the supplies of insurance intermediary services and not the significance of the income which the appellant derived from the relevant supplies;

we also do not agree that cases such as London Clubs, St Helen’s and Lokn’Store are authority for the proposition that the extent of the economic link between a cost and a supply is relevant to the question of whether there is a direct and immediate link between the two. We need to be quite clear about what we are saying in this context. As we observed in our preliminary observations at the start of this section, we are not saying that an economic link is irrelevant in considering whether there is a direct and immediate link between a cost and a supply at the attribution stage in the process of determining input tax recovery. We are simply saying that the answer to that question depends on the precise nature of the economic link. The economic link in one case might be direct and immediate and the economic link in another case might be indirect and yet, in both cases, the economic link could be equally substantial. In those circumstances, the former would give rise to an attribution of the cost to the relevant supply whilst the latter would not. In both cases, what matters at the attribution stage in the process is identifying the nature of the link (whether direct and immediate or indirect). Whilst economic analysis is likely to be involved in determining that nature, the extent of the financial consequences deriving from the link, and whether or not they are substantial, is not a relevant factor at that stage.

In this regard, it is important to bear in mind that the cases referred to above all related to the second stage in the process of determining input tax recovery – the apportionment stage - in circumstances where it was already accepted that the relevant costs either had a direct and immediate link with both taxable and exempt supplies or fell to be treated as overheads. That being so, the focus in those cases on the extent to which each category of supply made economic use of the relevant costs is not surprising. At the apportionment stage, the relative extent of economic usage will clearly be relevant. However, that does not mean that the extent of economic usage can somehow convert an indirect link into a direct and immediate link for the purposes of the first stage; and

in seeking to establish that there was a direct and immediate link between the cost of the adverts and the supplies of insurance intermediary services, Ms Barnes and Mr Reynolds - at least when they were making their primary argument - sought to rely on the fact that there was such a link between the cost of the adverts and the supplies of sofas and that the supplies of sofas and the supplies of insurance intermediary services were so closely-intertwined that that must mean that the same link existed between the cost of the adverts and the supplies of insurance intermediary services. We do not consider the closely-intertwined nature of the two categories of supplies to be of any significance in this regard. We agree with Ms Sloane that there was a similar link between the supplies of tickets and the supplies of catering in ROH and that this did not lead the Court of Appeal to conclude that the production costs – which had a direct and immediate link with the supplies of tickets – also had a direct and immediate link with the supplies of catering. A concerted add-on strategy does not create a direct and immediate link between the original cost and the supply of the add-on. Only the costs associated with the marketing and promotion of the add-on are directly and immediately linked with the supply of the add-on.

To adopt the analogy of a road journey, if:

a cost is incurred in order to reach destination A;

nearly everyone reaching destination A is sought to be persuaded to continue their journey on to destination B; and

the majority of travellers getting to destination A are in fact persuaded to continue their journey on to destination B,

then the link between the cost incurred in order to reach destination A and destination A is direct and immediate. Similarly, the link between the cost incurred at destination A in order to persuade the travellers to journey on to destination B and destination B is direct and immediate. However, the link between the cost incurred in order to reach destination A and destination B is only indirect. Travellers reaching destination B would not have done so but for incurring the cost of getting to destination A but that is not the relevant test for establishing whether a link is direct and immediate.

disposition

197. For the reasons which we have given, we uphold each Appellant’s appeal.

Right to apply for permission to appeal

198. This document contains full findings of fact and reasons for the decision. Any party dissatisfied with this decision has a right to apply for permission to appeal against it pursuant to Rule 39 of the Tribunal Procedure (First-tier Tribunal) (Tax Chamber) Rules 2009. The application must be received by this Tribunal not later than 56 days after this decision is sent to that party. The parties are referred to “Guidance to accompany a Decision from the First-tier Tribunal (Tax Chamber)” which accompanies and forms part of this decision notice.

TONY BEARE

TRIBUNAL JUDGE

Release date: 29 APRIL 2022