The UK Court of Appeal ("CA") has unanimously overturned the first instance finding of infringement in Interflora v Marks & Spencer and has ordered a retrial. The CA's decision to order a retrial is unusual, although it is most likely a function of the unique circumstances surrounding this litigation. The CA also reiterated that under UK (as well as under EU) law, internet advertising on the basis of keywords corresponding to trademarks is not inherently objectionable because its purpose is, in general, to offer internet users alternatives to the goods or services of the trademark proprietors. Keyword advertising will amount to trademark infringement only where it suggests an economic link with the trademark owner or is vague as to whether such a link exists.
This case concerns the use by Marks & Spencer ("M&S") of the word "Interflora" as an advertising keyword in Google's search engine. Searches involving the keyword led internet search engine users to advertisements for M&S flower delivery services. M&S had used the Google AdWords service, bidding on a number of keywords containing the mark "Interflora" and close variations of it. The relevant M&S advertisements appeared on a separate part of the search results known as "sponsored links" and did not feature the "Interflora" name.
The case has had a protracted history, which included a reference to the Court of Justice of the European Union ("CJEU") and a number of subsequent UK appellate rulings on evidentiary and procedural matters. The proceedings finally came to trial in 2012, and first instance judgment was delivered in May 2013. The trial judge, Arnold J, found M&S liable for trademark infringement.
Why Did the Court of Appeal Order a Retrial?
It is standard practice for the CA to determine the issue of infringement according to the correct legal principles where it sets aside a first instance ruling. In this case, however, the court said that whilst such an approach would have obvious benefits in terms of time and costs, it was not in a position to substitute its own findings because it had not been taken to all of the documents or had the benefit of hearing the oral evidence.
Whilst retrials in IP cases in England are rare, there are a number of factors that explains the CA's reluctance to determine infringement in this case. First, the evidence that was led at trial was both substantial and complex. Much of this evidence was not canvassed before the CA. Second, much of the witness evidence was impressionistic and/or was concerned with whether and to what extent there was consumer confusion. Given that the CA had not heard the oral evidence and was not in a position to assess witnesses' demeanour and approach, the cautiousness of the court is understandable. Finally, it is also important to recognise that the relevant legal tests in issue require a very nuanced and careful balancing of various factual considerations. Given the fundamental nature of the errors made by the trial judge, the CA simply believed that it was not in a position to determine how these issues would have been determined if the correct approach had been adopted.
When is Keyword Advertising Objectionable?
The CA were at pains to emphasise that using keywords that are identical to trademarks in online advertisements is not an inherently objectionable practice. Referring to earlier CJEU authority, the court reiterated that it will amount to infringement only where the advertisement does not enable the average consumer, or enables him only with difficulty, to ascertain whether the relevant goods or services originate from the trademark owner or an undertaking economically connected to it or, on the contrary, from a third party.
The practical effect of this legal formulation, however, is that it effectively requires establishing confusion even in cases involving identical marks and identical goods and services (so-called double identity infringement). This is not entirely straightforward given that liability for double identity infringement historically has been generally understood to be absolute, and proof of confusion is only part of the statutory wording where an identical or similar sign is used in relation to identical or similar goods or where a trademark has a reputation. The trial judge attempted to resolve this anomaly by finding that to the extent that likelihood of confusion needs to be shown in double identity cases, the onus of proof is reversed such that it is up to the alleged infringer to show that the advertisement is not objectionable. The CA rejected this reasoning. Although the advertiser needs to ensure that its use of keywords does not misleadingly suggest a link with the trademark proprietor, the legal burden of proving infringement lies entirely on the trademark owner.
The CA also rejected the trial judge's reliance on the recently developed concept of "initial interest confusion", namely where a consumer may be initially misled by use of a trademark, but that confusion is then dispelled before it can be acted upon. The court said that the doctrine, which is borrowed from US law, should not perform part of the analysis in claims of this nature as it is "unnecessary" and "potentially misleading".
Who Is the Average Consumer?
It is well established that the average consumer is a "reasonably well informed and reasonably observant internet user". The trial judge found that even though a majority of such persons would not have been confused with M&S's advertisement, a significant proportion of such consumers did believe there was a connection with M&S and Interflora. It was on this basis that he found M&S liable for infringement.
The CA held that a finding of infringement is not precluded by a finding that many consumers, of whom the average consumer is representative, would not be confused. In determining the issue, the court is faced with a binary question and not a statistical test. This is interesting because whilst the CA expressly recognised that survey evidence may sometimes be of assistance, such evidence will be of limited probative value in determining the approach of the hypothetical average consumer. Indeed, attempts by Interflora to introduce survey evidence were rejected by the CA twice in the course of this litigation itself.
The CA's decision to order a retrial in this case is unusual for litigation in England. However, the unique circumstances of the case mean that this is unlikely to set a general precedent for retrials of IP cases in England, even though the CA will never have had the benefit of seeing the oral evidence. The relevant factual matrix and the legal issues will be sufficiently clear in the vast majority of cases for the CA to properly dispose of the dispute.
The ruling also makes clear that keyword advertising will be objectionable only where it misleads (or is liable to mislead) the average consumer as to the origin of the goods or services. Given that even the trial judge acknowledged that the majority of internet users are now aware of the distinction between sponsored and natural search results, this will be an increasingly difficult threshold to satisfy.