Interflora Inc & Anor v Marks & Spencer Plc (Rev 1)

This is the latest battle in the long war of attrition between Interflora and Marks & Spencer (“M & S”). The background is set out in my judgment in an earlier battle at [2012] EWCA Civ 1501 [2013] ETMR 11 (“Interflora 1”). I will assume that anyone interested in this judgment either has or will read it. Suffice it to say that, M & S having successfully attacked the grant of permission to Interflora to adduce evidence from witnesses selected from a “witness selection programme,” Interflora counterattacked by seeking and obtaining permission to call evidence from witnesses who, they say, give evidence of confusion in the real world. On 21 February 2013 Arnold J granted that permission. His decision is at [2013] EWHC 273 (Ch). Normally this court would regard that as a discretionary case management decision; but Kitchin LJ recognised that it raised a question of principle. For that reason we gave permission to appeal. Because the trial was due to begin in mid-April the appeal was expedited. At the conclusion of the argument we announced that the appeal would be allowed. These are my reasons for joining in that course.

Interflora compiled a list of over 100,000 people who had either bought flowers through Interflora’s website or who had participated in a previous survey. From that potential pool of over 100,000 people, Interflora have managed to find 13 people who, they say, give evidence of real world confusion. It is the evidence of these 13 people on which Interfora wish to rely in support of their contention that the use of words such as “M & S Flowers” in an advertisement displayed as a sponsored link on a computer screen following a Google search for Interflora is insufficient to enable a reasonably well-informed and reasonably observant internet user to tell that the flower-delivery service offered by M & S does not originate from Interflora.

In Interflora 1 I quoted part of the decision of the CJEU at an earlier stage of the litigation. That court had said:

“49.

Indeed, if the referring court's assessments of the facts were to show that M & S's advertising, displayed in response to searches performed by internet users using the word 'Interflora', may lead those users to believe, incorrectly, that the flower-delivery service offered by M & S is part of Interflora's commercial network, it would have to be concluded that that advertising does not allow it to be determined whether M & S is a third party in relation to the proprietor of the trade mark or whether, on the contrary, it is economically linked to that proprietor. In those circumstances, the function of the INTERFLORA trade mark of indicating origin would be adversely affected.

50.

In that context, as has been observed at paragraph 44 of this judgment, the relevant public comprises reasonably well-informed and reasonably observant internet users. Therefore, the fact that some internet users may have had difficulty grasping that the service provided by M & S is independent from that of Interflora is not a sufficient basis for a finding that the function of indicating origin has been adversely affected.”

I also said (with the agreement of Hughes and Etherton LJJ) that the purpose of controlling evidence at the interim stage was:

“[70] … not merely to avoid irrelevant (i.e. inadmissible) evidence; it is also to avoid evidence which is unlikely to be of real value.” (Emphasis added)

“[76] … But the remaining objections still hold good: viz. that the selected witnesses are not (or at least cannot be shown to be) a fair sample of the class of reasonably well-informed and reasonably observant internet users, with the consequence that there is no ground for any extrapolation on a statistical basis, or on the basis of any mathematical or logical probability, of the views of the selected witnesses as representing the effect of the M & S advertisement on the hypothetical reasonably well-informed and reasonably observant internet user. If evidence of this kind cannot form the basis for extrapolation on the basis of any mathematical or logical probability leading to a conclusion about the effect of M & S's advertisement on the hypothetical reasonably well-informed and reasonably observant internet user, then in the absence of special circumstances it cannot be useful. And if it cannot be useful, it should not be allowed to distract the focus of the trial even if it is technically admissible.”

“[135] … Unless the court can be confident that the evidence of the selected witnesses can stand proxy for the persons or construct through whose perception the legal question is to be answered it simply represents the evidence of those individuals.”

“[142] … even at an interim stage a judge who is asked for permission to adduce such evidence should evaluate it carefully in order to see (a) whether it would be of real utility and (b) whether the likely utility of the evidence justifies the costs involved.” (Emphasis added)

“[143] … In the general run of cases it seems to me to place an undue and unfair burden on the other party for one party to tender in evidence witness statements from selected respondents to a questionnaire without even undertaking to produce a selection that demonstrates the full range of answers to the questions. As Mr Hobbs said, that places the burden on the defendant to disprove the validity of the selection, rather than on the claimant to validate it. Such a burden could in my judgment only be justified if the party tendering the evidence can show that it is likely to have a real impact on the outcome of the trial.” (Emphasis added)

“[144] … The current practice… is to allow the evidence in unless the judge can be satisfied that it will be valueless. In my judgment that is the wrong way round. I consider that, even if the evidence is technically admissible, the judge should not let it in unless (a) satisfied that it would be valuable and (b) that the likely utility of the evidence justifies the costs involved.”

“[146] In the present case I do not consider that Interflora has demonstrated that the evidence it wishes to call would be of real value. To put it bluntly, Interflora starts with an unreliable dataset from which it proposes to select the witnesses most favourable to itself. (Emphasis added)

[150] In deciding whether to give permission, the court must evaluate the results of whatever material is placed before it. Only if the court is satisfied that the evidence is likely to be of real value should permission be given.” (Emphasis added)

Mr Hobbs QC rightly said that this judgment was intended to send the general message that evidence from consumers in this kind of case (i.e. trade mark infringement involving ordinary consumer goods or services) should only be admitted if it is of real value; and even then only if the value justifies the cost; and that judges should be robust gatekeepers in that respect.

In Interflora 1 I had also said that “different considerations may come into play where … [e]vidence is called consisting of the spontaneous reactions of members of the relevant public to the allegedly infringing sign or advertisement”; and that “if there is evidence of consumers who have been confused in the real world, there can be no objection to calling it.” This appeal engages that aspect of Interflora 1.

The judge described the way in which the witnesses were identified. I need not repeat it. Once witnesses had agreed to participate in court proceedings, they were telephoned by solicitors representing Interflora who questioned them according to a script. Having noted their answers to the scripted questions, the solicitors then prepared witness statements which the witnesses signed. The three most relevant questions were:

“[26] What did you think when you saw M & S when you entered the search term “Interflora”?”

“[27] Why do you say that?”

“[28] From your memory of these search results, what, if anything, do you think the results tell you about any relationship between Interflora and M & S?”

Question [28] was to be asked only if the interviewee had not mentioned any connection between Interflora and M & S.

It is instructive to compare the raw answers to the questions with what ended up in the witness statements. The following table sets out the answers to those questions and the corresponding part of the witness statement. I have designated the witnesses by their initials; and have also indicated the question to which the answers were given.

The first point to make is that the witness statements paint a rather different picture to that given by the raw answers to the questions. The way the witness statements paint the picture is more favourable to Interflora’s case. They do not distinguish between the answers to questions [26] and [27] which might be described as the interviewees’ spontaneous reactions, and their answers to question [28] which in my judgment cannot. It is plain from looking at the answers themselves that many of the witnesses were led by that question into speculating about something they had not thought about before. The witness statements eliminate much of the witnesses’ hesitancy and uncertainty. They exclude key statements that some of the witnesses saw no connection between Interflora and M & S; and recognised that they were competitors. I do not consider that the witness statements are a fair reflection of the raw data. In my judgment, therefore, it is necessary to concentrate on the raw data rather than the witness statements.

What do these data really add up to? Witness (13) (IMF) can be eliminated immediately. The M & S advertisement had no impact on him. I note that his answer to the question was that he did not think anything of the relationship, but his witness statement said that he did not think much of it. The same, in my judgment goes for witness (1) (CRJ). She could see no relationship between Interflora and M & S, even after prompting, although she thought others might. Her answers to that effect found no place in her witness statement. Witness (2) (PH) had given no thought to the question before prompting. This was a field of speculation upon which she would never have embarked had the question not been put. Again, her witness statement made no mention of that. Her evidence was presented as if it were her spontaneous reaction. The same is true of witness (4) (JA) who assumed from the question itself that there must be some kind of relationship. This was not reflected in the witness statement. The same is true of witnesses (6) (JLM) and (10) (HP) who gave no thought to it at the time, but who were prompted to think about it by question [28].

So far as witness (7) (BA C-H) is concerned, her immediate reaction was that M & S and Interflora must be competitors. It was only after prompting that she speculated that that might not be the case. Her witness statement gives a quite different impression. That appears to me to have been the immediate reaction of witness (9) (NGJ) too. But his witness statement does not fairly reflect that. It gives the impression that he was confused at the time (although he himself did not use the word “confused” at all).

That leaves witnesses (5) (JS), (8) (LCT), (11) (GDF) and (12) (DA): four out of a pool of over 100,000. On the other hand there were 24 other interviewees who had no difficulty in separating M & S from Interflora. In those circumstances it seems to me to be difficult to say that the evidence of these four witnesses (or even all thirteen) demonstrates more than that some internet users may have had difficulty grasping that the service provided by M & S is independent from that of Interflora; which as the CJEU has said is not a sufficient basis for a finding that the function of indicating origin has been adversely affected.

In the present case, having considered the evidence Arnold J concluded as follows:

“[59] … Overall, I consider that this evidence is likely to be of some value in assisting the court to answer the question identified in paragraph 3 above. I do not think it is likely to be decisive, but nor does it appear to be of little or no value. How much weight to give it will be a matter for evaluation after the witnesses have been cross-examined and once it has been put in the context of all the other evidence available at trial.”

In reaching that conclusion the judge first held that the evidence was evidence of real world confusion. He quoted from the witness statements of four of the witnesses, although he did not at that stage refer to the raw data themselves. I regret to say that in my judgment the judge was misled by the impression created by the witness statements when compared to the raw data on which they are based. The judge said at [40]:

“Where there is evidence of actual confusion, however, there have been many cases in which the courts have found it of assistance to hear from people who were confused and who are prepared to come to court. Usually, it is quite hard to find people who have been confused, because confusion by its very nature often doesn't come to light, and it is even harder to persuade confused people to come to court to give evidence. Thus the available sample is often a small one. But experience shows that it is rarely completely repetitious. On the contrary, it often shows a range of responses to the sign in issue. In my view, it is helpful for the court to receive first hand evidence of the range of responses from the people who had those responses.”

One problem in this case is that the evidence of the 13 witnesses is not representative of the range of responses to the questions. Interflora acknowledge that; because they have disclosed the answers to the questionnaires of a further 24 interviewees who thought that there was no connection between Interflora and M & S. The judge brushed aside M & S’s complaint on that score. He said at [43]:

“Counsel for Marks & Spencer complained that no statements had been obtained from those 24, but I see no need for Interflora to have done so. If Marks & Spencer wish to obtain statements from those witnesses, their contact details have been provided.”

In my judgment, with all respect, this is an inadequate ground for dismissing the complaint. Part of the concern about the admission of this kind of evidence is that it is too partisan (or, as I put it in Interflora 1, it is evidence from a skewed selection of witnesses). In Interflora 1 I said at [143]:

“In the general run of cases it seems to me to place an undue and unfair burden on the other party for one party to tender in evidence witness statements from selected respondents to a questionnaire without even undertaking to produce a selection that demonstrates the full range of answers to the questions. As Mr Hobbs said, that places the burden on the defendant to disprove the validity of the selection, rather than on the claimant to validate it. Such a burden could in my judgment only be justified if the party tendering the evidence can show that it is likely to have a real impact on the outcome of the trial.”

In my judgment that is what the judge has done in this case. He has left it to M & S to amplify the range of responses to the questionnaire; and thus has imposed upon them the burden of disproving the validity of the selection. Mr Silverleaf submitted that any deficiencies in the witness statements could be explored and exposed in cross-examination for which Interflora had provided all the relevant material. But that, too, casts the burden on M & S to disprove the validity of the evidence rather than requiring Interflora to validate it at the “gatekeeping” stage.

The judge acknowledged that witness statements are rarely (if ever) the unmediated words of the witness himself. He continued at [49]:

“It is often been remarked that one of the problems with witness statements is that they often contain, in effect, the answers to a series of leading questions. In some cases judges have passed comments about witness statements being "heavily lawyered", meaning that the wording of the statement was more the product of the lawyer than of the witness. That is, of course, to be deprecated. Nevertheless, witness statements are, and have for some years been, an established feature of English civil litigation procedure. That being so, it would not be justified to require higher standards for the admission of consumer evidence in trade mark cases. That is particularly so where the gathering of the evidence has been made as transparent as it is here.”

It is not clear to me what the judge meant by referring to “higher standards” for the admission of consumer evidence in trade mark cases. As paragraph 18.1 of PD 32 makes clear, a witness statement “must, if practicable, be in the intended witness’s own words.” We know what the witnesses’ own words were: they are recorded in the answers to the questionnaire. It is not a question of higher standards: it is a question of ensuring that standards are complied with. If, therefore, the judge thought that he was being asked to impose higher standards, I respectfully think that he was wrong. In addition, if the evidence proposed to be called is relied on as consumers’ spontaneous reaction to an allegedly infringing sign, it is all the more important to eliminate answers to leading questions which almost by definition do not produce spontaneous reactions. This was at the heart of guideline (iv) of the Whitford guidelines (see Imperial Group plc v Philip Morris Ltd [1984] RPC 293 summarised in Interflora 1 at [61]):

“…the questions asked must not be leading; and must not direct the person answering the question into a field of speculation upon which that person would never have embarked had the question not been put.”

As the judge rightly said at [34] it does not matter whether, technically speaking, this is a survey or not. Whether or not it is a survey the value of the evidence is severely diminished, if not eliminated, by a failure to follow the Whitford guidelines. If compliance with the Whitford guidelines is a higher standard of evidence gathering from consumers, then that has been a feature of this kind of case for nearly thirty years. Whether or not this exercise is described as a survey, the answers to question [28] exhibit the vices of a poorly conducted survey.

The judge went on to say at [51]:

“On the application, Interflora is not seeking to put before the court the results of an experiment in the sense that I have just described. On the contrary, it is attempting to adduce evidence of what consumers’ responses were when presented with Marks & Spencer's advertisements when they were in the real world situation.”

It is, I accept, possible to describe the answers to questions [26] and [27] as the spontaneous reactions of members of the relevant public to the allegedly infringing sign, although even they were asked following a series of probing questions. But I do not consider that answers to question [28] can be described in that way. They are answers to a stimulus. Thus I do not agree that it is possible to describe the answers to question [28] as real world responses to the M & S advertisement. On the contrary, since question [28] was only to be asked if the interviewee failed to mention a connection arising out of his or her experience in the real world, it was plainly an attempt to elicit an answer by means of an artificial stimulus. In my judgment, the judge mischaracterised the evidence by labelling it evidence of “actual confusion” if, by that, he meant confusion in the real world.

The judge went on to consider M & S’s objection to the particular form of question [28]. He noted that Mr Hobbs QC had argued that it was essentially the same question as had been rejected at the previous stage; and acknowledged that that was “the strongest argument in favour of the exclusion of this evidence”. However, he was not persuaded by the argument for three reasons:

In my judgment the first of these reasons is wrong. Question 28 (to which the argument was directed) was not directed at the actual experience of the witnesses. That much is clear from the fact that it was only to be asked if the interviewee failed to mention a connection. The second reason acknowledges the vice of switching from the past to the present. But it underplays the obvious confusion between historic experience and response to a leading question which is inherent in the way that the question was framed. The vice of question [28] is seen most clearly in the answer given by JA: “I guess from your question, I assume there must be some kind of relationship.” I agree with Mr Hobbs that where one interviewee actually articulates the effect of the question, it is very likely that others either thought it; or did not realise the effect of the question on their answers. I have set out above a comparison of the evidence in the witness statements with the answers to questions on which that evidence is based. From that it can be seen that the evidence that Interflora wish to call is heavily dependent on the answers to question [28]. Without the answers to question [28] there is very little of it left. Finally the judge asked himself the question whether this meant that the evidence was of “little or no value”, a test which he amplified at [59] by saying that he considered that the evidence had “some value”.

With the benefit of hindsight, perhaps I did not make my message clear enough in Interflora 1. Let me say it again, but more loudly. A judge should not let in evidence of this kind unless the party seeking to call that evidence satisfies him (a) that it is likely to be of REAL value; and (b) that the likely value of the evidence justifies the cost.

Thus in my judgment the judge should have asked himself whether the evidence was likely to be of real value. The negative way of framing the test means that evidence which is not of real value will be admitted in a case which, after all, is about the provision of ordinary consumer services.

I am unable to tell from the judge’s statement that he considered that the evidence would be likely to have “some value” whether he simply thought that the value would be more than minimal or whether he thought that it would carry real weight. If the latter, I find it hard to identify why he thought that. In my judgment, therefore, the judge conducted a flawed analysis of the nature of the application, the quality of the raw data that supported the witness statement, and ultimately applied too lax a test.

I would also add this. In Interflora 1 I said at [147] that an application to adduce this kind of evidence should be made as early in the case as possible. The claim in this case was issued in 2008. The trial is listed in a two day window beginning on 15 April 2013. This application to adduce the evidence was made on 22 January 2013 and heard by the judge on 14 February 2013. The lateness of the application is particularly pertinent in the light of the judge’s rejection of M & S’s complaint that the selection of witnesses was partisan. As noted, the judge rejected that complaint on the basis that M & S had the contact details of 24 potential witnesses and it was up to them to decide whether or not to call them. In the light of the short time between the outcome of the application and the trial date (which the judge did not refer to at all) I regard that as placing a further unfair burden on M & S. Moreover, when he came to consider whether the value of the evidence justified the cost of the exercise he said that M & S had already spent the major part of the money needed to deal with the evidence. In so saying, the judge left out of account any further cost that might be incurred in locating, interviewing, taking witness statements from and calling the 24 interviewees whom Interflora had identified. This, too, vitiates the balancing exercise that he conducted.

As is well known the growing cost of civil litigation has caused extensive changes to the Civil Procedure Rules, which are due to come into force on 1 April 2013. Among the most significant changes is the recasting of the overriding objective so that it now becomes dealing with cases justly and at proportionate cost. This will make it all the more important for judges to exercise their power to limit or exclude technically admissible evidence which is not of real value.

For the reasons I have given I concluded that the judge’s discretion was exercised on a flawed basis, such that this court should exercise the discretion afresh. For the same reasons I am clear in my own mind that if I had been dealing with the application I would have refused it. Those were the reasons that led me to join in the decision to allow the appeal.

I agree. When I first read the papers in this appeal I was inclined to take the view that this was no more than a case-management decision with which we should not interfere. But Mr Hobbs’ direct focussing on the questionnaire and the witness statements convinced me that nothing of value could come out of the proposed evidence. Eight of the thirteen proposed witnesses had been exposed to the obviously unfair question [28]. The value of their evidence thereafter was obviously nil. Even if the remaining five stood up to cross-examination, one would be miles from being able to assess whether they were remotely representative of the public and so their evidence would be of no value.

This decision does not mean that the days of survey evidence are over. It is possible to conduct fair surveys and they may indeed lead to witnesses of value. And that may turn a case - some of the Jif lemon witnesses were of that character. But if the survey amounts to no more than scratching around for something and produces so little as this one has, then there is every good reason in common sense and procedural economy for excluding it along with any resultant witnesses.

I agree with both judgments.