Ruminations on the Likelihood of Consumer Confusion Standard in Trademark Law

By Eric Goldman

Last month, I attended the Third Trademark Scholars’ Roundtable in Bloomington, Indiana. See my prior blog posts about the first and second roundtables. See a photo of the participants. As usual, Rebecca acted as event chronicler. See her comprehensive blog posts (1, 2 (with artwork by Bill McGeveran!), 3, 4).

The roundtable’s theme was the likelihood of consumer confusion (LOCC) test and its deficiencies. Overall, the roundtable participants generally agreed that the test has obvious doctrinal deficiencies, but unfortunately we did not reach a consensus either on the underlying harms that trademark law should ameliorate or any fixes to the existing LOCC multi-factor test. I called on the group to come up with a consensus proposal, figuring that if our group of knowledgeable and largely like-minded academics could not come up with a consensus proposal, no one could. Sadly, we didn’t rise to that challenge.

I introduced one of the panels. The three main points I made in my introductory remarks:

1) Trademark law protects both producer and consumer interests. To the extent trademark law is protecting consumer interests, the LOCC test attempts to measure what’s going on in consumers’ heads (i.e., how the senior and junior users’ activities affect the consumers’ psychological processes). That inquiry is ultimately doomed. Consumer psychological processes are too complicated and heterogeneous for a legal test to accurately measure these effects.

2) The individual factors in the standard LOCC test are a weird mix of producer- and consumer-oriented factors. This mix of factors doesn’t make sense, which is why courts sometimes pull them apart and put them back together in interesting ways. (See the Network Automation case for a good example of remixed factors). Structurally, the LOCC test implicitly assumes that all consumers use the same search processes, but this assumption isn’t defensible. Different products have different search processes; and even for a single product, different consumer subcommunities will use different search processes.

3) Courts have developed a number of bypasses to the standard multi-factor LOCC test, especially to deal with cases involving non-competing trademark uses. Examples include initial interest confusion and post-sale confusion. However, these tests rarely improve the judge’s analysis; and as an empirical matter, I believe most courts reject the bypass tests and just go back to the standard LOCC test. For example, over the past few years, initial interest confusion gets mentioned fairly rarely in court opinions, and it has been dispositive in those cases even less frequently. I am planning to take a closer empirical look at citations to the initial interest confusion doctrine over the summer.

Having said all that, it’s my own ad hoc empirical observation that most courts reach the right result when using the LOCC test, even if the test (or the judge’s application) is flawed. The errors can be conspicuous, but most cases get to equitable results most of the time. The bigger issue are the errors in decision-making that never reach a court case, such as companies that steer away from grey areas in socially suboptimal ways or potential defendants who cave in response to a trademark owner’s C&D. (With respect to the latter, I still think a threats action may be useful).

Even so, I wonder if we could do simple things that would reduce the errors, reduce the adjudication costs, or empower potential defendants to make decisions with less uncertainty. It seemed like the LOCC test has been stretched to too many different types of TM owner-vs.-defendant factual positions. It is also conspicuous that the LOCC test is completely common law; Congress has never expressly adopted the test. Thus, I wonder if we might find it useful to statutorily adopt the test but then give explicit statutory guidance on how to apply the factors (just like the fair use test gives some explicit guidance to courts about applying the test, although we might view fair use as a cautionary tale rather than a model).

For example, we could statutorily introduce some “buckets” which give judges guidance about when to use the LOCC test or do something different. For example, I offered up the following four different buckets: (1) direct competitive uses, (2) TM uses in adjacent products, (3) merchandising/licensing, (4) everything else. In some sense, courts have already implicitly developed some different tests reflecting these different categories. For example, we have special rules for counterfeiting as a specific case under bucket #1; the courts do different things in merchandising cases; and the LOCC test works particularly poorly in bucket #4 (which is where I think the LOCC test errors are the greatest and most pernicious).

One other interesting takeaway I’ll mention here. At one point we talked about our frustration with the marketing literature that discusses trademark issues and how those papers often do not reflect a nuanced understanding of trademark law. At the same time, when we talk with marketing professors, they are baffled by how trademark law doesn’t reflect well-accepted insights in the marketing community. It seems that there could be some benefit to getting the trademark law professor and marketing professor community together to discuss how we can work together better. If you’re a marketing professor and might be interested in such an endeavor, please contact me.