Update to Barton Beebe's empirical study of TM

Kevin Blum, Ariel Fox, Christina J. Hayes, & James (Hanjun) Xu, Empirical Analysis of Multifactor Tests for Trademark Infringement

Interesting data from 15 years of SDNY trademark cases, coded and written up by four Harvard Law students. Singling out the bit on survey evidence:

Plaintiffs presented survey evidence in 24 of 139 preliminary injunction and bench trial opinions (17.3%), but this evidence was favorably credited by the court in only one-quarter, or six of these opinions. When the survey was credited, however, the court always found that the actual confusion factor favored the plaintiff and always ultimately found a likelihood of confusion (a plaintiff win rate of 100%). In the 18 opinions where the survey was not credited, none of the plaintiffs were able to win the actual confusion factor and only three were nevertheless able to obtain a finding of likelihood of confusion (a plaintiff win rate of 16.7%).

Despite the overall infrequency of opinions in which plaintiffs actually presented survey evidence, the court drew an adverse inference against the plaintiff for failure to submit survey evidence showing actual confusion in 13 opinions of our set of 139 preliminary injunction and bench trial opinions.. . . These data are particularly remarkable given that five of these 13 opinions were preliminary injunction opinions, when one would be less likely to expect the parties to be able to have had the time to prepare credible survey evidence.

Of the 13 cases in which the adverse inference was drawn, only one resulted in a finding that this factor favored a likelihood of confusion (7.7%), and only two opinions resulted in an overall finding of likelihood of confusion (a plaintiff win rate of 15.4%).

Defendants presented survey evidence in 13 of 139 preliminary injunction and bench trial opinions (9.4%), and the court favorably credited this evidence in six of the 13 opinions (46.2%). Defendants who presented credited survey evidence enjoyed a much higher likelihood of a finding of no likelihood of confusion, with only one of the six facing a finding of likelihood of confusion (a defendant win rate of 83%). Of the seven opinions that did not credit the defendants’ survey evidence, four resulted in a finding of likelihood of confusion (a defendant win rate of 57.1%).

My take: clients who can afford a survey are probably well advised to have one; though the risk of disregard is high, the payoff is substantial.