Alzheimer's Association and Alzheimer's Foundation in keyword battle

Among other things, this case has some interesting things to say about IIC and proper controls in survey cases.

Alzheimer’s Disease & Related Disorders Association, Inc. v. Alzheimer’s Foundation of America, Inc. 2018 WL 1918618, No. 10-CV-3314 (S.D.N.Y. Apr. 20, 2018)

The Association (counterclaim plaintiff) sued the Foundation (counterclaim defendant), alleging that the Foundation’s purchase of Asssociation trademarks as search engine keywords and use of the two-word name “Alzheimer’s Foundation” constituted trademark infringement and false designation of origin under the Lanham Act. The court found that confusion was unlikely.

The parties first litigated confusion in 2007, starting with fighting over checks made out to one entity but sent to the other.   The Association was formed in 1980 and is the world’s largest private non-profit funder of Alzheimer’s research. It has more than 80 local chapters across the nation providing services within each community. In fiscal year 2016, the Association raised more than $160 million in contributions and spent $133.6 million on program activities, including more than $44 million on public awareness and education. The Association had nearly 9 billion “media impressions” and more than 41 million website visits in that FY.  

In 2015, when survey respondents in the general population were prompted for the first two health charity organizations that come to mind, “respondents were vastly more likely to name the American Cancer Society (46%), St. Jude Children’s Research Hospital (28%), American Heart Association (19%), and Susan G. Komen for the Cure (10%), than they were to name the Alzheimer’s Association (3%).” When they were asked which organizations “involved in the fight against Alzheimer’s disease” they have heard of, the Association registered 8-12% awareness from 2011-2015. Among the demographic groups that the Association targets with its advertising messages, unaided awareness of the Association among organizations “involved in the fight against Alzheimer’s disease” hovered between 10% and 20% from 2009-2015. Aided awareness of “Alzheimer’s Association” was roughly 25-32% among the general population, and at 35-47% among targeted subgroups between 2011- 2015.  For the Foundation, those numbers were 15-19% and 25-32% respectively.

The Association has a standard character mark registration for ALZHEIMER’S ASSOCIATION, registered since June 8, 2004, but in use since 1988. The Association also has other registrations using “Alzheimer’s Association” along with other words or with graphical elements, as well as standard character marks for WALK TO END ALZHEIMER’S and MEMORY WALK. In 2016, nearly 500,000 participants took part in Association walks in 630 communities, raising more than $78.6 million. The Association websitet displays “alz.org” and “Alzheimer’s Association” at the top of its landing page. Its principal color is purple. 

The Alzheimer’s Foundation of America was founded in 2002 and has more than 2,600 member organizations throughout the country that collaborate on education, resources, best practices and advocacy. AFA has awarded millions of dollars in grant funding to its member organizations for services such as respite care. In 2010, AFA’s “revenues, gains and other support” from “contributions and special events including telethons” was approximately $6.6 million. Its website is at www.alzfdn.org, and principally uses the colors teal and white. Its first registered mark, from 2006, is for “AFA Alzheimer’s Foundation of America” with the organization’s “heart in hands” logo.

The Foundation also has a registration for “Alzheimer’s Foundation” plus the heart in hands logo.

The Foundation disclaimed any exclusive right to the literal elements of these marks.  In 2014, the Foundation filed for a standard character mark for “Alzheimer’s Foundation,” now registered on the Supplemental Register with a claimed first use date of 2004.  Before 2009, its use of the two-word name (without “of America”) was mostly limited to press releases and sponsored ads online, as well as on the Foundation’s Twitter account.  Since 2004, the Foundation has described itself in the header of online ads using the two-word name “Alzheimer’s Foundation,” resulting in over 30 million impressions between June 7, 2004 and the end of 2009. The Foundation also bought Association marks as keywords. From April 2012 to June 2014, the Foundation ran sponsored ads that used the word “Association” in the text, ending when the Association objected.

The Association also used keyword advertising, including buying “Alzheimer’s Foundation” as a keyword until 2010.

A Google search of “alzheimer’s association” from June 2014 showed the Foundation’s ad as the top result, with the main header as “Alzheimer’s Foundation - alzfdn.org” with the tagline “An Association of Care and Support. Reach Out to Us for Help....” The second and only other ad was from the Association. Their header reads “alz.org - Alzheimer’s Association,” and the tagline “Honor a Loved One with a Tribute Donation - Support Research & Care.”

“Alzheimer’s Association” sometimes led to more clicks on the Foundation’s sponsored ad than those received from its own brand. Indeed, campaigns targeting Foundation’s competitors performed the best and comprised roughly 40% of AFA’s keyword marketing budget.   During some of the relevant period, the Foundation may have been using Association-related metatags, though the court found this immaterial “as [site metatags] have likely not been used by search engines since before 2009, and so have little effect on the ordinary prudent consumer.”

On the Foundation’s donation page, there are many references to “AFA” or “Alzheimer’s Foundation of America,” and no references to the Association or use of any the Association’s marks. “At no point while on the AFA website during the donation process would a consumer see any of the Association Marks.” So too in reverse for the Association.

The Foundation was the first to complain of confusion, in 2004, when its then CEO wrote “a routine web search under ‘alzheimers foundation’ led me to [the Association’s] site. ‘Alzheimer’s Foundation of America’ is a registered service mark of our organization. It distresses me that supporters of our respective organizations may be confused when searching the internet.” This might not have related to sponsored ads, though. In 2014, a Foundation employee wrote that a survey showed many people saying they donated before or were “introduced to AFA through a fund raising event,” and speculated that “several respondents may have us confused with the Association or with ‘the cause.’ ”

Between 2007 and 2012, the Association received more than 5,700 checks made payable to “Alzheimer’s Foundation” or a variant totaling over $1.5 million. The Foundation received more than 5,000 checks between 2006 and June 2016 made payable to “Alzheimer’s Association” or near variants. A large percentage of the Foundation’s online donors are first-time donors, and that the average online donation, as well as check donation, is under $100.  The Foundation argued that the number of Foundation-labeled checks received by the Association was only 0.1% of the total number of checks received, and 0.252% of the total value of checks received.

There was evidence that a check for the Association was received alongside a printout of the Foundation’s internet donation form. When asked why she had submitted the Foundation form, the donor claimed that she had typed the Alzheimer’s Association name into her web browser and clicked on the first entry that came up to download the form for donations and that she was unaware that she was on the Foundation website.  No one testified about other donors or potential donors contacting one organization looking to donate to the other, apart from people asking about the difference between the Association and the Foundation.

Other purported instances of confusion included that NBC’s Today show ran a quotation the Association has provided alongside the Foundation’s logo, but that was the logo with the words “of America,” and wasn’t at issue in this litigation. In Celebrity Family Feud, the host once suggested the show was raising money for “Alzheimer’s Foundation,” when it was actually raising money for the Association. These instances didn’t have a clear link to the Foundation’s allegedly infringing actions.

There was also two studies from the Association and an expert critique by the Foundation.  Study 1 found 34% net confusion between the standard character marks “Alzheimer’s Association” and “Alzheimer’s Foundation.”  The court found that this was somewhat artificial but still probative to actual confusion. However, the court found that the control—“Alzheimer’s Trust”—artificially inflated the net confusion numbers.  They “pre-tested” two controls, “Alzheimer’s Charity” and “National Alzheimer’s Foundation,” which, by generating more confusion, would have yielded net confusion rates for Alzheimer’s Foundation of 12% and 11% respectively. 

The expert said “Alzheimer’s Charity” wasn’t good because it “sounded like a product category rather than a single real entity,” which sounds fair to me, but the court didn’t like the expert’s explanation overall.  He testified that his staff never disclosed the results of these pre-test surveys to him, even though the results were disclosed to the Association’s attorneys. “While it is legitimate to run a pre-test or pilot study for the purposes of improving a study, no credible explanation was offered for the changes made between the pre-test survey and the reported survey, and this suggests a potentially improper purpose.” 

Setting that aside, “Alzheimer’s Trust” was a weak control term. Though testing a two-work mark made sense, “Trust” as a descriptor for a charity was both more unique than “Foundation,” and more easily distinguished from “Association” in part because of its multiple meanings. “Alzheimer’s Federation” could have more clearly controlled for the confusion created by reasons other than Foundation-specific reasons, or “Alzheimer’s Foundation of America,” given that the Association was arguing that it was the two-word mark that was confusing.  “[W]hat better way to test that proposition than to compare its use with that of AFA’s full, and undisputedly non-infringing name?”

Another study allegedly showed IIC.  Respondents were asked to type “Alzheimer’s Association” into a search box and then were either shown results including the Foundation’s disputed ads and other keyword ads (test condition), or organic search results without ads (control condition). Respondents were asked to click on the link/s they thought would take them to the website of the organization for which they searched. If they clicked on the Foundation’s ad, they saw the Foundation’s web page and were asked if they thought it was the web page of the organization for which they had searched. If respondents didn’t select the “Alzheimer’s Foundation” or control links, respondents were asked to click on the link or links, if any, that they believe would take them to the website of an organization affiliated with the organization they searched for.

There was a net rate of 20% sponsorship confusion, and, of the 42% who were confused in the test condition, 74% remained confused as to source or affiliation after viewing a screenshot of the Foundation’s web page (31% of the total respondents in the test condition).

The court found it significant that the study design assumed that participants asked to type in “Alzheimer’s Association” knew that it is a particular organization. “Consumers cannot mistakenly associate the AFA ad or website with the Association if they were not aware of the Association’s existence to begin with.”  The test stimulus was also biased—there was no sponsored ad for the Association, whereas the Association also bid on its own marks as keywords, and often would have appeared in the sponsored ads section as well.  The order of the ads would have changed based on the organizations’ respective bids, and thus order bias was also implicated—a bias that couldn’t be accounted for by the chosen control.  [citing Winnie Hung, Limiting Initial Interest Confusion Claims in Keyword Advertising, 27 Berkeley Tech. L.J. 647, 666 (2012) (stating that statistics show ad rank has a significant effect on consumers’ confusion).] 

The court, however, concluded that though the bias suggested a lower actual confusion percentage, the study did use a real web page with real ads that did actually result from typing in “Alzheimer’s Association” at one point in time. “[I]f the Court were to find a likelihood of confusion based on the one test stimulus used, it would constitute a Lanham Act violation just the same.”  [Is that really likely confusion if there weren’t evidence about how often it would have happened?]  

The study inflated net confusion because the first search result in the control was the Association’s website, while the first search result in the test condition was the allegedly infringing Foundation ad.  “A better control stimulus would have contained non-infringing sponsored ads at the top that, if clicked, counted towards the confusion rate for the control.” What was at issue was the specific ads, not sponsored ads in general, and thus “only the offending item should have been removed or replaced.”

Overall, the evidence “strongly militates against placing much weight on the study evidence.”

The Association’s evidence of intentional attempts to confuse also faltered.  Of possible interest:  “AFA’s patent counsel filed two false specimens of use”—in its filing for a standard character mark, t counsel included as a specimen of use a page from its magazine in which the words “of America” had been obscured. After the Association raised the issue, counsel filed press releases as substitute specimens of use, which were accepted by the USPTO.  The Foundation also filed a section 8 declaration with respect to the composite mark and included one specimen of use (among others) in which “of America” was removed and that had not actually been used in commerce. The Foundation again filed an amendment after the Association raised the issue, and the USPTO accepted the amendment.  There was no evidence that the Foundation was aware of either of the false specimens of use.  [Still, looks bad for the lawyers. To the extent that this increases the expense/burden of litigation, it may raise questions that should be taken up with an insurer.]

Because the Association didn’t challenge “Alzheimer’s Foundation of America” in any context, the key question was the [marginal] likelihood of confusion arising from the use of the two-word name and the Foundation’s use of Association marks in search keyword ads and metagags. 

[One way to read this case might be as a cautionary tale for the assumption that structures much discussion of the TM/unfair competition interface—that there is something that can really be done to limit confusion without disallowing use of generic terms, or functional features, or whatever isn’t protectable by TM but may still be contributing to consumer confusion.  Then-Judge Ginsburg might have wanted there to be a good solution for the two Blinded Veterans association, but we have no empirical evidence that any intervention would work.]

[Also, another interesting question here is the appropriate comparator keyword buy.  Suppose that the Foundation had merely bought “Alzheimer’s.”  Its ads should then have stilled displayed in response to a search for “Alzheimer’s Association,” though possibly at a different price point depending on the search engines’ practices. The consumer doesn’t know why the ad displays.  If that’s the appropriate comparator, then the net confusion from buying the Association’s trademark would be zero.  Or should we also allow the Association’s trademark rights to force the Foundation to use negative keywords, as 1-800-Contacts made its competitors do (in violation of antitrust law, as the ALJ found)?]

Anyhow, the court noted that it wasn’t evaluating the Foundation’s keyword buys in a vacuum, but rather the effect of the keyword purchases in conjunction with the Foundation’s resulting ads.  Further, the key type of confusion was IIC.  Was the Foundation engaging in a bait-and-switch likely to confuse consumers, or offering consumers a choice?

Though the Association’s mark was incontestable, it was still commercially weak, which was crucial to the court’s holding here.  As the Ninth Circuit has said, “a consumer searching for a generic term is more likely to be searching for a product category. That consumer is more likely to expect to encounter links and advertisements from a variety of sources.”  Both components of the Association’s mark were descriptive of the relevant charitable endeavor. Althought the Association is the world’s largest Alzheimer’s-related non-profit and the world’s largest non-governmental funder of Alzheimer’s research, the consumer studies in evidence showed that the mark’s secondary meaning was not strong.  Between 10-20% unaided awareness among potential donors wasn’t much.

“Here, because of the weakness of the mark, it is not easy to disaggregate the consumers searching for the Association as a specific organization from those searching generically for an Alzheimer’s charity when they type ‘Alzheimer’s Association’ or some similar derivation into a web browser.” Relatedly, it was difficult to distinguish consumers confused by the Foundation’s actions from those “confused simply by the similarity of the descriptive marks.”

Mark similarity: the court reasoned that the proper comparison was not the mark to the keyword buy itself, but the mark to the resulting ads, because that’s how consumer confusion would manifest itself.  The Association argued that, in organic search results, AFA’s website appears as “Alzheimer’s Foundation of America,” and that the use of the two-word name in the ad header and sometimes “association” in the ad text combined with the keyword buys to increase the similarity between the marks/ads. 

The Foundation argued that, in context, the similarity was “mitigated by the unique information and URL of its ads,” but the URLs www.alz.org and www.alzfdn.org and ad text, e.g., “Honor a Loved One with a Tribute Donation. Support Research and Care.” and “An Organization Providing Support for Your Loved One with Alzheimer’s” weren’t particularly dissimilar. And differences in the parties’ websites came too late.  The ads were similar in appearance and meaning, favoring the Association.

The Foundation tired to argue that it was primarily focused on providing care services while the Association focuses on funding research, but that didn’t distinguish them in the market for donors.

Actual confusion was a key factor, but the evidence “must be related to the actions or behaviors at issue.”  The evidence of 11,000 mislabeled or misdirected checks was probative of confusion “generally,” but not by confusion generated by the Foundation’s actions.  Nor were the limited instances of confusion in the media and anecdotal reports of consumer confusion persuasively related to the Foundation’s actions. “In light of the descriptive nature of the marks, confusion between the unchallenged ‘Alzheimer’s Foundation of America’ mark and ‘Alzheimer’s Association’ supports the notion that it is the weakness of the marks and consumers’ inattention, not AFA’s specific disputed practices, that yields confusion.”  

Finally, the surveys’ flaws meant they merited little weight.  The point of the surveys ought to have been to pin down causation by focusing on the challenged behaviors, but that was exactly what they did not do.

Intent: The Association argued that the Foundation complained of the confusion created by the similarity in the organization’s names back in 2004, but, despite its own complaints and lawsuits, then proceeded to employ Association marks in its metatags and as keywords, as well as attempting to register a two-word mark without really using it.  The Foundation argued that it didn’t intentionally use false specimens and that its internet ads had been similar since 2004.  The court didn’t find the Foundation’s trademark counsel “particularly credible in his explanation of the false filings, but circumstantial evidence about AFA’s trademark applications does not itself suggest an intention to capitalize on the Association’s goodwill or to exploit confusion.”  And the Foundation noted that the Association also bought Foundation keywords before 2010, and coexisted with the Foundation and its practices for six years before suing.  When the Association complained about the Foundation’s use of the word “association” in its ad text in 2014, the Foundation removed the offending ads.

The court found that intent favored the Foundation.  “In many respects, the lack of direct evidence of actual confusion undermines the Association’s attempts to argue that AFA acted with the intention of exploiting consumer confusion.” The evidence didn’t support the claim that the Foundation believed the specific practices at issue here were causing consumer confusion. Because this factor failed, the Association also failed to show the intentional deception vital to IIC.

Consumer sophistication: the relevant population was the average consumer searching for the Association online and seeing the Foundation’s ad, who are less likely to be institutional donors and many of whom are first-time donors. “[I]n this day and age, the ability to complete a form on a website does not itself make consumers particularly sophisticated.” This factor favored the Association.

Weighing: the most important factors here were strength of the mark, similarity of the marks, and evidence of actual confusion, two out of three of which strongly favored the Foundation.  A final consideration didn’t neatly fit within the Polaroid factors: “the labeling and segregation of online advertising.” The “ad” label “heightens consumers’ care and attention in clicking on the links, and further diminishes the likelihood of initial interest confusion.”  [Or more likely, contributes to ad blindness if it’s noticed at all.]

Highlighting my ongoing conviction that elaborating causation stories makes actionable confusion less likely to be found, the court elaborated that it was trying to figure out how many consumers fell into the relevant subpopulations: (1) Those who weren’t confused, either because they were using “Alzheimer’s Association” in a generic sense in their search or because, on seeing the search results, they understood the difference between the two organizations. (2) Those who clicked on the Foundation’s ad by mistake and who were diverted because of the Foundation’s keyword buys and its use of the two-word name “Alzheimer’s Foundation.”  A subset of (2) could remain confused even after viewing the Foundation’s website.  (3) Those who were mistakenly diverted for other reasons, such as because of the inherently weak and descriptive nature of the parties’ marks. “These consumers are ‘confused’ in the colloquial sense, but would have been confused even if they searched the word ‘Alzheimer’s’ alone or even if AFA solely utilized its full name.”  In light of the evidence, it was difficult to identify the proportion of consumers in each group, and thus the Association didn’t show a probability, as opposed to a possibility, of confusion.

Finally, and perhaps exhausted by the effort so far, the court rejected the Association’s attempt to cancel the Foundation’s marks.  As to one mark, the court accepted the dubious rationale that filing a Section 8 declaration in 2017 showed lack of abandonment.  Even though the court didn’t see where the challenged mark, as opposed to the other mark, was included as a specimen of use in that filing, “the filing of the application alone demonstrates AFA’s intent to use the ‘014 Mark. Given the high burden placed on the Association to establish abandonment, and the limited evidence adduced, the Court cannot find abandonment.”  I can’t see how that can be right—if the challenged mark isn’t in the filing, given multiple years of alleged nonuse, that can’t be the end of the analysis, unless there’s some sort of implicit tacking analysis going on.