Consumer surveys as evidence of trade mark infringement

By Katharine Stephens


This article was written by IP partner Katharine Stephens and it was first published by Intellectual Property Magazine, October Edition (see PDF version here).

Visit IP magazine at

Master of all we survey?

As the UK’s courts and Intellectual Property Office subject consumer survey evidence to significant controls, Bird & Bird’s Katharine Stephens looks at how their use varies between other EU countries.

Consumer survey evidence is useful in a number of instances. A claimant in a trademark case may wish to prove that there is a likelihood of confusion between its mark and the sign that the defendant is using, but for one reason or another, perhaps because the defendant’s activities are not wide spread, the
claimant does not have any evidence of confusion. Alternatively, a claimant may wish to prove the strength or fame of its mark, or a defendant may want to prove that the mark has become generic. In such instances, one or other party may want to prove the point by use of a survey of the relevant public.

For the purposes of this article and with the help of our colleagues, we looked at the issues governing the use of survey evidence in Belgium, Germany, the Netherlands, Sweden, Spain, Italy and Finland in comparison to the UK. Despite the harmonised nature of trademark law, when and how consumer surveys can be used by litigants to best effect varies from country to country in Europe.

What are surveys used for?

In the UK, survey evidence is often used to demonstrate a likelihood of confusion in a trademark action or that there is a misrepresentation or a likelihood of deception in a passing off action. For example, survey evidence was successfully used to support Associated Newspapers’ claim that the publication of a paper called Evening Mail would cause confusion with their titles, Daily Mail and The Mail on Sunday [1]. However, the use of surveys in the UK is not limited to the issue of confusion; it can also be adduced to show that a mark has acquired distinctive character as in the “Viennetta” case where Unilever sought, but failed, to show that the shape of their ice cream was distinctive [2].

In contrast, the German courts consider confusion to be a point of law rather than a point of fact, meaning that survey evidence cannot be used to show a likelihood of confusion. Survey evidence in Germany is therefore used to demonstrate:

• acquired distinctiveness, with respect to the earlier mark, in opposition proceedings;
• increased distinctiveness and notoriety in infringement actions; and
• consumer understanding of a product’s position and reputation in the market.

In practice, strong evidence of confusion may still be persuasive in building up an overall impression and is often submitted to help tip the balance whilst not being acknowledged as such.

The courts in Belgium, the Netherlands, Sweden, Italy and Finland do not limit the use of surveys. The courts will allow survey evidence to be used to show likelihood confusion, although in Sweden it is reasonably rare, and also to demonstrate a mark’s reputation or distinctiveness, or to show detriment to the reputation of a mark.

In Spain, the use of surveys is generally confined to showing the well-known character or reputation of a trademark.

Survey as a witness collection programme

A survey, in the strict sense, is one where a whole range of views is collected and then collated, analysed and subjected to statistical analysis, such as when demonstrating the level of confusion present in the hypothetical average consumer. This is how they are conducted and the results presented in civil law jurisdictions.

In the UK, surveys are not only presented as statistical reports, but are also conducted to find witnesses to appear at trial. Surveys used as witness collection exercises are often not conducted or presented to the courts as surveys in the strict sense. However, they are still considered to be surveys by the court and subject to the same rules governing such evidence [3]. As one Judge noted, re-labelling a survey as a witness collection exercise does not improve its quality [4]. The evidence from witnesses obtained by such means has been relied upon even where the survey itself was not or was conceded not to have been statistically valid, or where the questions were acknowledged to be flawed. For example, in Numatic International v Qualtex5, a passing off case relating to the shape of the claimant’s ‘Henry’ vacuum cleaner, the Judge relied upon the evidence of the witnesses identified during the survey, despite the survey questions being misleading. It is a question of degree as to whether a witness produced as a result of a flawed survey is likely, through no fault of his own, to give misleading and tainted evidence. A poorly conducted survey can result in the witness evidence being excluded as happened in UK Channel Management.

Best practice

In most jurisdictions, the courts have commented in varying degrees about the best practices for conducting consumer surveys. In the UK, the basic parameters were set out in Imperial Group v Philip Morris [1984] RPC 293: “If a survey is to have validity (a) the interviewees must be selected so as to represent a relevant cross-section of the public, (b) the size must be statistically significant, (c) it must be conducted fairly, (d) all the surveys carried out must be disclosed including the number carried out, how they were conducted, and the totality of the persons involved, (e) the totality of the answers given must be disclosed and made available to the defendant, (f) the questions must not be leading nor should they lead the person answering into a field of speculation he would never have embarked upon had the question not been put, (h) the exact answers and not some abbreviated form must be recorded, (i) the instructions to the interviewers as to how to carry out the survey must be disclosed and (j) where the answers are coded for computer input, the coding instructions must be disclosed.”

In Finland, the National Board of Patents and Registration outline on their website the requirements which must be followed for surveys used in applications for trademarks with a reputation. These requirements are similar to those noted above in Imperial Group and have the practical effect of requiring a party to disclose the full survey report in unabridged form. Although these requirements apply to a small proportion of Finnish trademark surveys, they can be viewed as guidance of what the courts will expect in all surveys.

Sweden’s Court of Patent Appeals, which is the court of appeal for administrative trademark proceedings, has also given very similar guidance. In most other jurisdictions, there is no requirement for the party who conducts the survey to disclose anything more than the results. Certainly, in Germany, a party would expect to disclose only the statistical results, and not answers in full. However, in many other jurisdictions, this sort of information is almost always disclosed to prevent the other party challenging the reliability of the survey and to ensure that the survey appears as reliable and impartial as possible. Where, as often happens, both parties undertake their own surveys which, unsurprisingly, tend to give conflicting results, a well thought out and robust methodology, as
well as disclosing as much background information as possible, is key if a party wishes to maximise the chance of their results being adopted by the court.

A further piece of guidance, highlighted particularly by our colleagues in the Netherlands, Belgium, Finland and Germany, is that surveys should be performed by an impartial expert at a reputable market research company. If an impartial expert is not used, the courts are likely to give less weight to the results. In contrast, in the UK, trainee solicitors at a party’s law firm can be used to conduct a survey provided they do not have any involvement in the dispute [6].

Pitfalls to avoid

Even well designed surveys can go wrong, for example, a question or series of questions that is thought to be clear is misunderstood by the interviewee. However, common mistakes include:

The survey is not representative: Interviewees should be selected so as to represent a cross-section of the relevant public. In Dualit v Rowlett
Catering Appliances [7], the survey designed to show that the shape of a toaster was distinctive was flawed because it started with a filter question relating to income; it was limited to those who were more likely to be buyers of Dualit’s expensive products.

The questions are leading or cause unfamiliar speculation: Judges have expressed concern that questions should not lead the interviewee to the answer that the party wants. In some cases, it is easier to spot a misleading question than in others. In Interflora v Marks & Spencer [8], the Judge refused to allow witness evidence following on from a survey as the negative emphasis in the phrasing of the question implied, contrary to fact, that there was a connection between Marks & Spencer and Interflora. The offending question asked, “thinking specifically about this search result (points to Marks & Spencer result), is there anything that tells you it is not related to Interflora?”

Relatively benign looking questions have also been criticised. For example, in The Rugby Football Union v Cotton Traders (an infringement case relating to the rose device used by the English rugby team), the Judge criticised the use of the word “logo” in questions such as “what does this logo mean to you?” because some of the witnesses understood the word “logo” to mean “trademark”.

The interviewee should not be directed into a field of speculation upon which he or she would never otherwise have embarked. In Specsavers International Healthcare v Asda, the Court of Appeal upheld the Judge’s finding that not much weight could be attributed to the survey. The survey was conducted near an Asda store and the question “if I wanted to buy a pair of glasses made by Specsavers, do you know where is the nearest place that I could get them?” may well have prompted some of the interviewees simply to guess that Specsavers’ glasses were available in Asda.

Omnibus surveys: The court has been unimpressed with omnibus surveys (surveys which related to several different products). In UK Channel Management, the Judge observed that the survey suffered from a number of evidential difficulties. Because the claimant was not the sole person who paid for questions, it could not produce the questionnaire or the instructions to interviewers and did not have the answers given by the interviewees. Consequently, it was of such little value that the Judge did not allow it to be adduced.

The courts in the Netherlands and Sweden have also commented on poor practices. Both courts, unsurprisingly, take a negative view on leading questions. The Dutch courts have also indicated that they are likely to disregard survey evidence if it does not accurately reflect the factual situation, does not represent the relevant period of time or is not conducted in a sufficiently objective manner. The Swedish courts believe that a methodology is poor if the questions asked do not reflect the purpose of the survey and are not relevant to the facts at issue. They have also commented negatively about surveys where the selection of candidates is too narrow, meaning the survey does not capture a representative sample of the ‘relevant public’. The Swedish courts are particularly wary if this is as a result of surveying in too small a geographical area.

Supervision of surveys

The UK is unique amongst the jurisdictions in that survey evidence is only admissible with the court’s permission. The rationale for the strict control was explained in A & E Television Networks. It ensures the evidence is admissible, probative and proportionate, whether it be the actual answers to questions, the expert evidence adduced to explain the results of the survey or the evidence from the witnesses collected as part of the exercise. It also prevents a party seeking to call a witness whose evidence was tainted by the mechanism under which was collected. Finally, it ensures that costs are not wasted and are proportionate. This approach has also been adopted by the UK Intellectual Property Office in Tribunal Practice Notice 2/2012. As noted above, such controls can result in survey evidence and witnesses identified as part of a survey being excluded as happened to one of the surveys, but not the other, in Interflora. However, on a more positive note, the Judge in A & E Television Networks saw it as an opportunity to allow the applicant to modify the survey following
contributions or objections from both the court and the opposing party.

In Germany, Belgium and the Netherlands it is possible, although rare, for the courts to independently order survey evidence to answer a point of contention between the parties. The Courts’ views of survey evidence ‘Scepticism’ is a word that recurs when we asked our colleagues how the courts viewed survey evidence where it was intended to support a case of confusion. In Sweden, it is rare for the court to rely upon such evidence and in the Netherlands, Italy and Belgium, Judges will tend to give such evidence the weight they feel it is due and will rarely take the results at face value. In the Netherlands, the courts go further and, in the situation where two surveys contradict, they will generally ignore the results from both. In Spain, the courts generally regard surveys to
support confusion as biased; therefore, in practice, they are rarely used for this purpose.

The position is slightly different where a survey is not being used to show confusion. For example, in Germany and Sweden, when surveys are used to provide evidence of fame or acquired distinctiveness, the courts tend to rely on the submissions of the parties when forming a view, although they will only be given significant weight where the courts believe that they have been conducted properly and the results are reliable.

In the UK, the weight which the Judges give survey evidence can be limited because the question of whether there is confusion, or the extent of a party’s reputation, is a matter for the court to decide and the results of the survey cannot alone be determinative of such issues. Furthermore, oral evidence from witnesses is frequently felt to be superior. In the ‘Henry’ vacuum cleaner case, the Judge noted that he would accept the survey as showing strong recognition of the ‘Henry’, but he preferred to rely on the oral evidence of the witnesses when it came to the issue of whether that recognition was sufficiently strong to conclude that purchasers would be misled.

As noted above, flaws in the way the survey is conducted can be fatal. However, if the survey supports the Judge’s opinion, the criticisms levelled at it are quite often overridden. For example, in Associated Newspapers the Judge said, “notwithstanding the defendant’s criticisms, the survey does give a feel for the reaction of members of the public to the defendant’s proposed newspaper … The outcome of the survey accords with one’s expectations.”

Final thought

There is a one important point which is shared between all jurisdictions and which we have not touched on already: cost. No matter what the jurisdiction, a survey is a time consuming and costly exercise. Consideration therefore should be given at the outset as to how to conduct the survey and whether the costs outweigh the benefits. Increasingly, however, survey evidence is being used in trademark litigation, despite the criticisms levelled at it and, in the UK, the controls placed upon it.


1. Associated Newspapers v Express Newspapers [2003] EWHC 1322 (Ch).
2. Unilever’s trademark applications [2002] EWHC 2709 (Ch).
3. Specsavers International Healthcare v Asda [2010] EWHC 1497 (Pat).
4. UK Channel Management v E! Entertainment [2007] EWCH 2339 (Ch)
5. [2010] EWHC 1237.
6. A & E Television Networks v Discovery Communications Europe [2011]
EWHC 1038.
7. [1999] RPC 890.
8. [2012] EWHC 1722 (Ch).
9. [2002] EWHC 467.
10. [2012] EWCA Civ 24.