Fage UK Ltd & Anor v Chobani UK Ltd & Anor

The Defendants in this action for passing off apply for permission to adduce survey evidence and thereafter evidence from respondents to that survey.

The Defendants have already conducted a pilot survey. The results are in evidence. The question now is whether a further survey is likely to be of real value, sufficient to justify the additional costs of the trial if survey evidence is admitted.

The Claimants oppose the application. They contend that the pilot survey, and the fuller survey proposed, are unreliable and valueless, and that a witness collection exercise based on the survey and results would in turn be so tainted as to yield no evidence of real utility to the court. They also submit that it is in any event premature to permit the Defendants to call evidence from respondents to the proposed survey since the Defendants are obviously not yet in a position to demonstrate how the witnesses were selected nor what they are likely to say.

The Claimants do not themselves seek to adduce any such evidence; and under directions made by Briggs J on 1^st November 2012 to enable a speedy trial the parties were required to apply to adduce such evidence by no later than 30 November 2012, and only the Defendants did so.

The main action is based on the extended form of passing off initially recognised in J. Bollinger v Costa Brava Wine Co Ltd [1960] RPC 16. That case concerned Champagne; this concerns Greek yoghurt.

The Claimants are producers of a brand of yoghurt sold as TOTAL Greek Yoghurt. As set out in their Skeleton Argument, the Claimants’ case at trial will be that (1) a valuable reputation and goodwill exists in the name “Greek Yoghurt” as denoting a particular type of yoghurt coming from Greece; (2) the Claimants own a share in that goodwill; and (3) that goodwill will be damaged by the Defendants’ sale of a product which is falsely described by that name because its product is not made in Greece but in the USA. Put shortly: only yoghurt made in Greece from Greek milk can be called Greek Yoghurt, just as only sparkling wine made in a particular way in the Champagne districts of France can be called Champagne.

The Defendants are producers of a brand of yoghurt called CHOBANI Greek Yoghurt, which they introduced into the UK market in September 2012. They contend that the yoghurt product they sell can properly be called “Greek Yoghurt”: they say that Greek Yoghurt connotes a yoghurt with a particular consistency and taste but does not denote that the product is made in Greece from Greek milk. An analogy might be English mustard, the characteristic being colour, taste and consistency, not origin.

The trial is listed to commence on the first available date after 18 February 2013. This has necessitated tight time-tables for pre-trial processes to be completed.

The Defendants have given undertakings suitably to re-label their products until after the hearing.

Until the very recent decision of the Court of Appeal in Marks and Spencer PLC v Interflora Inc and another [2012] EWCA Civ 1501 the practice was to allow such survey evidence in unless the judge was satisfied that it would be valueless.

However, in that case, the Court of Appeal considered that approach to be the wrong way round. Lewison LJ (at paragraph 144) stated that

Further, the Court of Appeal made clear that the decision should be made at an interlocutory stage, and that the judge faced with an application to let in such evidence should resist the natural temptation to leave questions of admissibility to trial, and determine the issue by balancing the cost of a survey against its likely utility.

Interflora was a trade mark infringement case: there was no claim in passing-off. Lewison LJ recognised that a survey and witness collection evidence may have more utility in the context of passing-off cases, where the question is whether a substantial proportion of the public has been misled and the test differs from the “average consumer” test in trademark infringement cases. Nevertheless Counsel were agreed that Lewison LJ’s prescription of a revised approach and procedure would generally be of application in both contexts.

To facilitate a decision at that early stage the Court of Appeal specified a new standard form of order to make it clear (see paragraph 149) that

Lewison LJ also prescribed the following:

I turn first to the survey that the Defendants now seek to carry out further to the pilot survey they have already made.

The Claimants submitted first that the Defendants had not identified any pleaded issue that the survey truly addressed and should be disallowed accordingly.

I do not accept that: it seems to me that the issue as to whether a substantial proportion of consumers or potential consumers of Greek yoghurt would regard a fundamental characteristic or attribute of “Greek yoghurt” as being that the yoghurt must have been made in Greece with Greek milk, so that yoghurt manufactured elsewhere cannot be Greek yoghurt, is at the heart of the action. A survey properly designed and administered could provide useful evidence as to that fundamental ingredient of the Claimants’ case. The fact that the Defendants, rather than the Claimants, have taken up the task does not diminish its relevance or potential utility.

The second objection raised by the Claimants is of more substance. This is their contention that the survey as actually proposed is so flawed that it yields neither reliable statistical evidence nor a proper basis for further witness evidence and would be a waste of time and money. They based this on two principal arguments.

First, they submitted that the Defendants are in effect striving for the impossible: to prove a negative (that a substantial part of the relevant public would not be misled) from a limited sample size (of 1,000 selected persons pursuant to the pilot survey). I do not accept this argument.

It seems to me that in reality the Defendants seek to contradict the Claimants’ case (which it is, of course for the Claimants to prove) that consumers of yoghurt believe that Greek yoghurt is a badge or indicium which means that the produce to which it relates is made in Greece. In my view, a representative sample of 1,000 persons could be sufficient, if (a) calculated and likely to be truly representative of a relevant cross-section of the public and (b) approached and questioned in the course of an approach compliant with what are often referred to as “the Whitford guidelines.”

Secondly, and in that context, the Claimants submitted that both the general approach adopted by the Defendants and the specific questions proposed to be asked in the course of the survey are subject to such serious flaws as to undermine the reliability of the survey and render useless its results in evidential terms.

In support of these submissions the Claimants (1) criticised the way in which the survey had been set up (and especially the fact that the creation and conduct of the survey were designed by the Defendants’ legal team apparently without any input from a dedicated expert in the design and implementation of such surveys and (2) themselves backed up these criticisms with a Witness Statement dated 4 December 2012 from one of the best known such experts, Mr Philip Ian Malivoire (“Mr Malivoire”). Mr Malivoire gave evidence, and his experience was commended, in the Interflora case.

I address first the Claimants (and Mr Malivoire’s) criticisms of the approach suggested, which they suggest does not comply with the Whitford guidelines.

Mr Malivoire summarises his criticisms in the penultimate numbered paragraph of his witness statement as follows: “In the limited time I have had to contemplate the proposed survey and the broader context of the legal proceedings, my main observations are that, in my opinion, the methodology of the pilot survey conducted / proposed survey to be conducted is not sufficiently explained to draw reasonable conclusions as to how useful the results are likely to be. In particular, I have fundamental concerns over whether the methodology proposed will direct the questions to the relevant public; question 1 is, in my view, demonstrably too imprecise. In addition, there is, in my view, inadequate information as to how respondents are to be shown the product, too little information about the respondents’ characteristics (age and work status, for example) being gathered, and little or no detail as to what geographical locations are to be used for the full survey."

Taking each of these in turn, Mr Malivoire’s criticism of question 1, and his conclusion that it is demonstrably too imprecise, has two facets: (i) first, he suggests that inadequate definition is brought to the stage directions included in the first question that a FAGE yoghurt pot is to be shown and handed to respondent and (ii) secondly, that the first question is too vague and uncertain, and too open to interpretation, to act as an accurate filter for the purposes of assembling a representative cohort of representative respondents.

As to (i), Mr Malivoire fixes on the importance of clarity and consistency in the way that respondents are introduced to the pot of yoghurt about which they will be asked questions. He says that it is not clear, for example, whether they are to be allowed freely to inspect the pot or just be shown it, and he refers to the fact that one draft of the proposed survey did not provide for respondents to be shown the pot whereas others did.

He also makes the subsidiary point that it is not explained in the draft survey how the words “Produced in Greece” were and are to be masked. He submits that without knowing precisely how respondents will be shown or allowed to handle the product it is impossible to draw any conclusions as to the appropriateness of the methodology of either the pilot or the fuller proposed survey.

I can fully understand the need for precision and consistency in this regard. However, possibly in light of Mr Malivoire’s strictures, the Defendants have now clarified that respondents are first to be shown and then handed a FAGE yoghurt pot, and that the words “Produced in Greece” will be masked in a way that was demonstrated and seemed to me discreet and effective. I do not read Mr Malivoire’s evidence as suggesting that any of these steps is flawed if consistently followed. Accordingly, I do not think this criticism has any longer any real substance.

As to (ii), the first question is undoubtedly of particular importance in providing the basis later for selection of representative respondent; and it is not easy to frame so as to be quite clear without leading the respondent. The first of the Whitford guidelines is that “if a survey is to have any validity at all, the way in which the interviewees are selected must be established as being done by a method such that a relevant cross-section of the public is interviewed.”

The first question in the questionnaire is to read “Do you eat products like this…?” Mr Malivoire puts his criticism simply: he says it is not clear what “products like this” means: it might mean a dairy product, a yoghurt product generally, or a more specialist yoghurt product; and the range of potential answers might vary depending on what size of pot (they vary) is being shown. He draws attention to the fact that the flaws as he sees them are exacerbated by the binary nature of the question, with no indication how the interviewer may react to and record any request for clarification such as “What do you mean by products like this?”.

Further, Mr Malivoire points out that the question seems to eliminate respondents who do not eat FAGE TOTAL Greek Yoghurt (or “products like it”), but who may buy it or who may otherwise be aware of it: and it includes people who eat other yoghurt but would never eat Greek Yoghurt. He concludes from this that there is a risk that the survey may include non-relevant members of the public and exclude relevant members.

Lastly, he points out that the record of responses to the first question in the pilot survey shows that approximately 50% answered in the affirmative, which seems surprisingly high given that it seems unlikely that 50% of the UK adult population eats such yoghurt and suggests that the initial question is not adequately filtering out non-relevant respondents.

Again, I can see some force both in Mr Malivoire’s criticisms and in his implicit suggestions for improvement (for example by including provision for recording clarificatory questions and answers). I would myself tend to think that the first question might usefully extend to those who buy as well as those who eat “products like this” and perhaps for the word “yoghurt” to precede the words “products like this”; but, as is obvious, I have no experience of devising such surveys, and my preferences may well be subject to criticisms also.

Mr Malivoire also points out that “it is best practice to record at least the sex, age and work status of all respondents… and to set quotas based on these characteristics to ensure that the cohort assembled is representative.” He also says that the geographical spread of 6 locations in England is not justified nor likely to be sufficient (he would himself suggest in the region of 29-40).

Turning to the remaining specific questions proposed, Mr Malivoire’s overall criticism is that they tend either to invite speculation or encourage respondents to “give more thought to the question than they otherwise might under spontaneous circumstances” and he points to the consequent risk of them “second guessing what they might otherwise expect the commonsense response to be.”

More particularly, Mr Malivoire points to the use in question 2 of the phrase “do you know…” as it “risks masking respondents who think they know the answer but are wary of being too definitive (particularly under survey conditions)”; and he suggests that the phrase “does it matter” in each of questions 4 and 6 and “Where would you expect” in question 7 may have a similar effect, and invites speculation, contrary to one of the Whitford guidelines.

Survey evidence is notoriously difficult to design and conduct. As stated at paragraph 10-026 in Wadlow ‘The Law of Passing-Off” 4 ^th ed.

Mr Malevoire himself will be acquainted with the enormous difficulty of designing and conducting a survey: in the Interflora case Lewison LJ refers to having declined to admit one survey constructed by Mr Malivoire and having been sceptical of another second survey Mr Malivoire had provided thereafter in the context of UK Channel Management Limited v E! Entertainment Television Inc [2007] EWHC 2339 (Ch). No construct is likely to be beyond, and few if any are likely to survive, criticism.

It is for that principal reason that it is most unusual for the evidence to be confined to, or even principally focused on, surveys: the real purpose of surveys is to provide the means of selecting live witnesses who can be presented as (in a passing-off case) a representative proportion of the public or (in a trade mark case) satisfactory proxies for the legal construct of the reasonably well informed and reasonably observant consumer. But even then the reliability of the survey, in terms both of its conduct and its questions, is of obvious importance: a flawed survey is likely to lead to a skewed process of witness collection.

As Lewison LJ put it in Interflora at paragraph 64, if the survey is flawed and the rationale for its construction opaque “A cynic might think that the phrase “witness collection programme” is simply a euphemism for adducing evidence from a skewed collection of witnesses identified by means of a statistically invalid and unreliable survey.”

In short, though the survey will not stand alone the Court must be satisfied at the first stage of permitting it that both in terms of its construction and implementation and in terms of the actual questions asked it is likely to result in useful evidence.

However, as Lewison LJ recognised in paragraph 35 of his judgment in Interflora, “there can be no doubt that a valid survey can be an accurate diagnostic or predictive tool.” That, it seems to me is especially so in passing-off cases, and where the survey is directed to the first of the 3 hurdles in the classic three-part test (requiring a claimant to establish a reputation or goodwill in some badge or indicium of trade such that the badge or indicium is recognised by the public as distinctive specifically of the claimant’s goods). A survey and its product may be of real use in guiding the court as to the likely view of a substantial proportion of relevant consumers.

In approaching its difficult task the court is, to my mind, necessarily to take into account that reliability will always be subject to further argument, and all surveys are imperfect. There is a risk if the threshold is set too high that a potentially beneficial, even if imperfect, filtering process will be thrown away.

The difficulty of assessing whether the survey is good enough to pass muster, and the validity of objections to it, is the greater where, as here, one side has provided expert (even if expressly not definitive) evidence (that is in the witness statement of Mr Malivoire for the Claimants) and the other has not (the Defendants having relied only on Ms Mikheev). It is not easy to dismiss as insufficient (or as Counsel for the Defendants urged “nit-picking”) even the admittedly provisional views of a respected expert in a field where practical experience seems to me of especial importance.

Further, where what is now called for is (to quote Lewison LJ in Interflora at (paragraph 147) a “definitive ruling one way or another” the risk of injustice if the court proceeds at this stage with only one side supported by expert evidence is obviously the greater. A cost/benefit analysis is an invidious task in such circumstances.

As to the benefit of the survey as such, the results of the pilot survey must be seen subject to the qualifications and criticisms adumbrated above. Further, I consider that the fact that the pilot survey was carried out in a single location does tend to undermine its reliability as an indication of what a fuller survey might reveal. Even so, on balance, those replies are such as, to my mind, to suggest that a full survey could (a) assist in establishing whether the first hurdle of the three in a passing-off case is surmounted, and whether a substantial proportion of relevant consumers would believe that Greek Yoghurt is a badge or indicium which means that the product to which the expression relates is made in Greece.

As to the costs, those presently envisaged by the Defendants are comparatively, not to say surprisingly, modest. The estimated cost of the pilot survey is stated to be some £1,800 (allowing for the difficulty in dividing time between survey preparation and legal work; and an estimate of £10,000 is given by Ms Mikheev for a third party marketing company to conduct the survey.

But, of course, that does not include the Claimants’ consequential costs; nor the almost inevitable further cost if (as also seems almost inevitable) the Defendants have to engage an expert to defend the survey and any subsequent witness collection programme (if allowed) from the criticisms already foreshadowed by Mr Malivoire (from whom no doubt more would be heard also).

In other words, if the process now is permitted, the costs will in all probability substantially exceed the present estimates. By how much it is very difficult to tell on the present state of the evidence.

Meanwhile, the already tight timetable ticks on: there is little room for a further delay to enable imperfections to be addressed.

In this difficult situation, and with reservations that may appear obvious, I have eventually concluded that I should proceed as follows, recognising with not a little anxiety that in the particular circumstances I consider some departure from the usual course (now prescribed):

As is implicit in these directions I consider that a survey along the lines proposed may be of assistance in enabling relevant evidence to be adduced, but I do consider it likely that the questionnaire would benefit from expert input and entirely dispassionate consideration of the points raised by Mr Malivoire. But for the pressure of time (in the context of the directions made for a speedy trial). I would have been minded to decline to admit this survey and ask for further and better survey to be provided.

It does seem to me that the Court is in difficulties in dealing fairly and definitively with the ultimate issue as to what evidence it will permit in at Trial when it has expert evidence from only one side as to the reliability of the process and the cohort selected and the representative nature (or not) of the respondents’ replies. No doubt the Defendants will bear this in mind at the next stage of the process, especially since if (as they urge) such evidence should be permitted to be adduced, there will almost certainly be expert controversy as to its reliability.