The Court of Appeal, Civil Division, set out the principles and approach to be observed in proceedings alleging trademark infringement, when a proprietor of the trademark was seeking to rely upon evidence of witnesses identified by means of a witness-gathering exercise, where the proprietor had not intended to rely on the survey or questionnaire by means of which the witnesses had been identified.
The claimant entities owned the trademark 'Interflora'. They operated the largest flower delivery network in the world, including an online flower delivery service. The defendant company was a well-known national retailer. It operated a flower delivery service via its website. The defendant was not a member of the Interflora network, however, the defendant had purchased and used the keyword 'Interflora' through the online search engine facility 'AdWords'.
The result of that was that if an internet user searched for the word 'Interflora' using a Google search, a result for the defendant would be found along with a link to the defendant's website. That gave rise to the instant proceedings in which the claimants alleged that the defendant's use of that keyword through AdWords had amounted to an infringement of the claimants' trademark. The claimants had conducted exercises with a market researcher to carry out screening of potential participants who had purchased or looked into the purchase of flowers.
The participants to the exercises were asked a series of structured questions in regard to searches and search results carried out using 'Interflora' and the results that had brought up the defendant's pages. The claimants disclosed to the defendant the answers given by all the participants to the questions they were asked during the course of the exercises. The claimants accepted that the surveys were not statistically reliable, but proposed to call as witnesses those participants whose answers to the questions had been the most favourable to their case.
The judge granted the claimants permission to call those witnesses identified as a result of the first exercise on the ground that the evidence obtained from the participants had been admissible and might be probative even though the surveys themselves could not be relied upon. The defendant appealed.
The defendant submitted that the court should not permit evidence to be adduced from witnesses selected as a result of a survey unless the survey itself was statistically reliable. Further, inter alia, the question at issue was whether the reasonably well-informed and reasonably observant internet user would or would not have understood the defendant's advertisement to indicate that it was part of the claimants' network, that the reasonably well-informed and reasonably observant internet user was not a real person, but a legal construct, and, accordingly, simply to call some internet users to give evidence was not probative of the issue in the case. Consideration was given to the principles set out in Imperial Group plc v Philip Morris Ltd ( RPC 293) (the Whitford principles).
The court ruled:
(1) The current practice, which was to allow the evidence to be adduced unless the judge could be satisfied that it would be valueless was the wrong way round. Even if the evidence was technically admissible, the judge should not allow it unless he was satisfied that it would be valuable and that the likely utility of the evidence justified the costs involved (see , - of the judgment).
What was in issue was the effect on a hypothetical legal construct, namely, the reasonably well-informed and reasonably observant internet user. Unless the court could be confident that the evidence of the selected witnesses could stand proxy for the persons or construct through whose perception the legal question was to be answered, it simply represented the evidence of those individuals. In a case in which the witnesses were called in order to amplify the results of a statistically reliable survey, their evidence might be probative. However, unless the court could extrapolate from their evidence, it was not probative. Where witnesses had been identified by a tailored series of questions, generally they would have been led towards a particular mindset which no longer represented the unstimulated evidence of people in the real world. That was not to say that there could never be evidence called in a case of trademark infringement. The court might need to be informed of, inter alia, shopping habits, the market in which certain goods or services were supplied, or the means by which goods or services were marketed. In addition, however, different considerations might apply where: (i) evidence was called consisting of the spontaneous reactions of members of the relevant public to the allegedly infringing sign or advertisement; (ii) evidence from customers was called in order to amplify the results of a reliable survey; (iii) the goods or services in question were not goods or services supplied to ordinary consumers and were unlikely to have been within the judge's experience; (iv) the issue was whether a registered mark had acquired distinctiveness; or (v) where the cause of action was in passing off, which required a different legal question to be answered. Accordingly, it could not be said that evidence from participants to a questionnaire could never be called in the absence of a statistically valid and reliable survey. However, the cases in which that kind of evidence might be of real use were difficult to imagine. Therefore, such evidence was not inadmissible as a matter of law. Even at an interim stage, a judge who was asked for permission to adduce such evidence should evaluate it carefully in order to see whether it would be of real utility and whether the likely utility of the evidence justified the costs involved (see , ,  of the judgment)
In the instant case, the judge had erred. The claimants had not demonstrated that the evidence which it had wished to call would have been of real value. They had started with an unreliable dataset from which they had proposed to select the witnesses most favourable to them (see , , of the judgment).
Accordingly, the appeal would be allowed (see  of the judgment).
Bach Flower Remedies Ltd v Healing Herbs Ltd  All ER (D) 1142 considered; UK Channel Management Ltd v E! Entertainment Inc  All ER (D) 149 (Oct) disapproved; A & E Television Networks LLC v Discovery Communications Europe Ltd  All ER (D) 34 (May) disapproved.
(2) The court could not make any order without some material on which to base its decision. Thus, there could be no objection to the carrying out of a true pilot survey, at the risk as to costs of the party carrying it out, before applying for permission to adduce the results of a survey. But that pilot survey would be no more than a basis for a further survey. Further, an application to admit survey evidence or evidence from respondents to a survey (or pilot survey) should be made as early as possible in the course of case management. It would not be right to leave it to the time when witness statements were exchanged. The objective of such an application was to have a definitive ruling one way or the other. It was a natural temptation for a judge who was not immersed in the case to leave questions of admissibility to trial. But balancing the cost of a survey (or witness collection exercise) against its likely utility, that temptation should be resisted. For the future, the standard form of order should be redrafted so as to make it clear that: (i) a party could conduct a true pilot survey without permission, but at his own risk as to costs; (ii) no further survey could be conducted or adduced in evidence without the court's permission; and (iii) no party could adduce evidence from respondents to any survey without the court's permission. In deciding whether to give permission, the court could evaluate the results of whatever material was placed before it. Only if the court was satisfied that the
evidence was likely to be of real value should permission be given. The reliability of the survey was likely to play an important part in that evaluation. Even then, the court could be satisfied that the value justifies the cost. That required the court to conduct a cost/benefit analysis. In a case of trade mark infringement in which the issue was one of deception in relation to the provision of ordinary consumer goods or services, those criteria were likely to be satisfied only in a special or unusual case. If what was sought was permission to carry out a survey, the applicant should provide the court with: (i) the results of any pilot survey; (ii) evidence that any further survey would comply with the Whitford guidelines; and (iii) the cost of carrying out the pilot survey and the estimated cost of carrying out the further survey. If what was sought was permission to call witnesses who had responded to a survey or other experiment, the applicant should: (i) provide the court with witness statements from the witnesses proposed to be called; (ii) demonstrate that their evidence would be of real value in deciding the issues the court had to decide; (iii) identify the survey or other experiment and, in the case of the administration of a questionnaire, disclose how many surveys had been carried out, exactly how those surveys were conducted and the totality of the number of persons involved and their answers to all questions posed; (iv) disclose how the proposed witnesses were selected from among the respondents to the survey; and (v) provide the court with the cost of carrying out the pilot survey and the estimated cost of carrying out any further work in relation to those witnesses (see - of the judgment).
Imperial Group Ltd v Philip Morris & Co  RPC 293 applied.
Decision of Arnold J  All ER (D) 57 (Jul) Reversed.