Legal Applications of Marketing Theory, part 4

Legal Applications of Marketing Theory, part 4

Steve Ansolbahahere & Jacob Gersen, Harvard University, Dept of Government & Harvard Law School, Consumer Confusion in the Law of Food (Are People Misled?)

Pom Wonderful case: Pom Wonderful Pomegranate Blueberry 100% juice; 85% pomegranate juice and 15% juice from concentrate.  Minute Maid made a Pomegranate Blueberry blend of 5 juices with almost no pomegranate juice.  Kathleen Sullivan for Coca Cola said that consumers weren’t so unintelligent not to recognize the blend; Kennedy says “I thought that this was pomegranate juice.” 

So what are people thinking, and in particular what do they want (what are their most material preferences)?  Nelson’s theory of search, experience, and credence claims.

Survey about which drink would be sweeter: POM 19%, Minute Maid 43%  [remainder: no difference]

Tastes better: 30% versus 28%

Costs more: 51% versus 15%

Is organic: 29% versus 7%

Is a Pomegranate drink: 54% POM, 10% minute maid

Regression: which would you buy?  Real pomegranate juice matters; taste is important; nutrition is important

Survey about role of price versus health versus taste.  Labels affect perceptions of various attributes; filtered through value or weight that people place on each attribute. Health and nutrition are usually what people care about most, so inferences about those things have very large effects.

Health claims about foods are tightly regulated; consumers are trying to get information that they can’t get from the info so they are making lots of cross-attribute inferences.

Halo/horn effects: inferences across all dimensions at once come from the label GMO: people assume that it’s as bad as trans fat, as bad as high calorie.

Q: Pom established perception of health w/juice.  Effect could be coming from the color of the juice, even from Minute Maid.  It wouldn’t be linear.

A: probably true.

Q: consumer preference not to be fooled?

A: there probably is—Scroogled campaign was effective in accusing Google of malfeasance.

Q: represented Ocean Spray in similar case. It wasn’t just the color; the labels had pomegranates all over them and not the juices that made up the bulk of the product.  Pomegranate was a hot ingredient! But the health point is important. The attributes aren’t just correlated but in some sense the same; healthy is highly regulated so sellers have found lots of other ways to say it. Any claim about presence/absence of any nutrient is interpreted by consumers as a general health claim.  So are you just getting at “health”?

A: there were pomegranates on both products’ labels, but health/nutrition perceptions differed between Pom and Minute Maid.

Q: but it tracked the health conclusions.

A: mindful of what is activating those impressions—it’s not consistent or easy.

Q: preference for truth speaks to a remedy, not necessarily a consumer preference. If I’m misled about something not material to my choice, maybe I don’t deserve $, but the labeler should still possibly be punished for lying.  Distinction b/t info given to consumer/respondent and actual label—when I rely on a label for health related information, I actively search for it. Color isn’t something I actively search for—I receive it passively. 

A: we varied things like calorie content, sugar, in our tests.

Q: but drawing their attention to it may make a difference.

A: he’s skeptical that people don’t look at labels.  A lot of people do, not every single time, but at least at purchase initiation.  If there was no benefit to saying it, the company wouldn’t spend money to say it. They think it matters.

RT: I would have some Qs about the role of the TM versus the specific juice, including what they think “pomegranate drink” means.  It could be just a brand effect/people thinking “pomegranate” means Pom Wonderful.  If that’s true then the regulatory challenge is even greater.

A: Doesn’t seem to happen with Wesson v. Mazola oil, or Muir Glen v. Swanson canned tomatoes.  [Which might be perfectly consistent w/a brand effect to the extent that Minute Maid and Pom are known for particular slices of the juice market, whereas those brands don’t have distinctive health/nutrition profiles in their categories.]

Discussion re: harm to consumer v. harm to competitor being different things.

David Hosp & Mark Puzella, Orrick & Orrick, Profit Disgorgement in Trademark Litigation

Needs to be tied to the relevance of causation.  Came out of work for Wal-Mart on Wal-Mart’s litigation over Backyard for grills (v. Backyard Barbecue for another store).  Found to have been willful infringement.  Now it’s a disgorgement case: but whatever Wal-Mart sells, it sells $1 billion—over $1 billion of grills/grilling accessories. The Lanham Act allows profit disgorgement. How do you fight out a damages battle where the damages might even be trebled [but it couldn’t go over $1 billion—it has to be compensatory and not a penalty, not like antitrust]. The judge was thinking $500 million.  We tried it and got it down to about $35 million.  Reversed on appeal; tried to jury and got $90 million verdict, now on appeal.  Comes down to the Q of damages definition.  Lanham Act doesn’t define infringer’s profits; case law is over the map.  Damages have to be attributable to the infringement.  P’s burden: show revenues. D’s burden: show what shouldn’t be counted, including costs of goods sold.  If you start with $1 billion, that might get you down to $450 million—now what?

Figure out what’s attributable to the brand.  If you’re looking for a $25 grill, nobody cares it’s called Backyard.  Different possible surveys, regression analysis.  Juries understand hard numbers.  Ultimately, P’s burden of showing revenues, but attributability gives D a chance to attack causation. The burden is on D to show lack of causation, but that still ought to be open as an avenue for the D to show it’s not tied to the infringement.

When he was starting out, people didn’t think TM cases involved damage awards. That’s changing, particularly as patent law is being curtailed in its competitive uses.  Companies are shifting to trade dress claims.  Mahindra: offroad vehicle they’ve been making for 70 years, initially under license from the Jeep corp. Sued by Fiat/Chrysler for trade dress infringement. Causation will get a lot of attention under the next few years.

Q: Going to defend the jury.  You use names in your presentation for authority—doesn’t that indicate something about the value of names?  Maybe that’s the role of the jury: to think about how they do their shopping and whether it matters.

A: We want to figure out the appropriate structure for showing that names matter.  [Or trade dress.]  Our chapter seeks a framework.  The jury needs instructions.

Q: Backyard v. something else. What’s the counterfactual? Wal-Mart not selling grills, or using the name Frontyard, or Weber, or something else?

A: Wal-Mart took Backyard off its products; used same labels, color scheme, etc. except it had no name whatsoever.  “Grill” instead of “Backyard Grill.” No impact on sales/sales went up slightly.  We would view that as the counterfactual.

Q: would you advise your client next time to roll out different styles in different places to establish the counterfactual?

A: we deal with that a lot.  W/a large corporation, you have to assume that the profit margin on white label is so much higher than national brands that companies are moving more and more into white label. They like having a brand name; they want it to be descriptive; inevitably someone has a registration for something that’s at least borderline close. [See Barton Beebe & Jeanne Fromer’s empirical work proving this.] That is a recipe for getting sued.

Q: why isn’t a company as big as Wal-Mart doing small experiments to measure possible damages?

A: this is a newer issue; we’ll see more companies hit with large verdicts.  There are also internal pressures from marketing folks who fell in love with a name and got it cleared.  Legal knows it can’t be the department that always says no.

Q: (1) what survey form/controls did you end up using that are public? (2) hypothesis: causation is important and underthought at this stage b/c modern TM lacks a materiality requirement in the first place, which it should have preserved from the old common law. (3) Also, be interested to have you speak to relationship b/t irreparable harm and difficulty calculating damages—does showing that it’s really hard to trace damages support the pre-eBay practice of having injunctive relief be standard?

A: survey asked people their motivations for purchase. Need more work on surveys about causal relationships with sales—may see more of those in survey world.

Q: survey world can definitely produce! Why did Wal-Mart choose a name if it can sell grills w/o?

A: consumers expect to see a name. Marketers want something that actually fits w/the product category, doesn’t turn the consumer off. Doesn’t have to drive sales, just make sense to the consumer.  Interesting Q whether it is actually necessary.

Q: causality argument on disgorgement makes it a mirror image of a lost profits claim. If the P can seek either form of remedy, is there a concern that by pushing causality into disgorgement we’re taking away one of the remedies the legislator thought was important?

A: depends on the case.  Maybe there is a name that really does drive sales.  Reverse confusion case.  Damages and profits are related, for sure. They should relate to one another.

Q: how much is a conceptual challenge v. challenge of proof?  Conceptually it’s clear that the damages are either lost profits or disgorgement of d’s profits.

A: there is a conceptual challenge b/c the language of the Lanham Act allows for both [if not duplicative].  There were instructions on both and awards for both in our case.  Notion of causation is not very well explored in the case law.

August Horvath, Foley Hoag, Damages Estimation in Consumer Deception Class Actions:  Legal and Methodological Issues

Chapter is about consumer class action damages: there’s a lot of blame to be spread around on judges, litigators, expert witnesses, system of litigation structure.  It’s unrealistic to say there’s no such thing as a model that can estimate classwide damages—judges will reject it because it will remove the remedy [I’m not so sure about that]. 

In most jurisdictions, the only way to show damage is to show price premium; the other possible ways to show harm don’t work in American courts.  What happened hypothetically in a world without the false claim?  Consider: powdered infant formula with powerful brand + false claim about preventing allergies.  Creates a new demand curve compared to the equilibrium pre-false statement.  Doesn’t affect the supply curve at all, since falsity is costless (unlike incorporating a patented feature into a product). Conventionally there should be a volume increase and a price increase.  Companies don’t really know their own demand curve (or supply curve), though, which complicates things.  Executive may choose b/t trying price increase and looking for volume increase.  In theory, false advertising damages to consumers are from just that change in price and supply.

But conventional damages models for class actions look at WTP, at consumers rather than at the consumer/producer interface. The basic Lanham Act damages measure is profits/damages diverted.  Does conjoint analysis get us to the actual surplus affected by shifting the demand curve? No, it doesn’t.  No conjoint analyst has ever convinced him that it explains the interaction between supply and demand, or whether the price produced by conjoint analysis lives anywhere on a supply/demand curve.  In principle, a part worth is individual to each consumer—some value an attribute negatively, positively, not at all.  But experts assume there is a part worth for each attribute representing WTP.  If the distribution is normal instead of uniform, then it’s not clear the result is the same.  If the center is the mean, then raising the price by that amount would seem to lose half the consumers (the attribute wouldn’t be worth paying that much for the half below the mean), so seller won’t do it. 

The supply curve is also relevant.  Some experts then assume that they’d raise prices so that the supply would be the same and the price would be higher. But that’s a completely unrealistic assumption about pricing. Courts have accepted it b/c they’re desperate.  Supply side: the words courts use to explain constraints on manufacturer other than demand—but it’s not quite the right words because of the presence of retailers/middlemen.  Wal-Mart has price requirements; the formula maker had no ability to raise prices at will.  At most it can take the increased sales at the same price.  Need more sophistication but at the same time learn the sloganeering that helps explain them to courts to avoid the bad analyses that courts have accepted in the past few years.

Q: vast majority of conjoint analysis he’s seen in litigation are crap b/c experts who produce them don’t understand it.  There are perhaps a dozen people who could do that right; his guess is that Horvath hasn’t seen those.  The supply/demand curves presented reflect monopolies, not competitive markets.  If Minute Maid has a false advertising campaign changing the probability of choice, competitors may or may not react; if you recognize a competitive market, the best way to explain it is not by adhering to a curve like that.  The best way to analyze it is a market simulation with a variety of competitors. If you change an attribute in conjoint analysis, you change a market share that is simulated by, say, 1000 respondents.  Need experts who understand the power of the tool.

Q: Some of these could be argued as estimates of the maximum possible change—the jury could see that simply.

Q: Conjoint is sensitive to so many small things; uniformity in design is lacking, even before you run any simulation.

A: at the level of class certification, that’s not where we need to focus—that’s for later in the analysis.  That kind of objection will be dismissed by the court as weight v. admissibility—the court just wants to know whether the model in principle can do the job at this stage.

Michel Pham, Columbia Business School, A Consumer Psych Perspective on Source Identification and Confusion

Consumer source identification—similar to identification of painter of a painting.  How does it work? Long-term memory & knowledge.  For source identification to take place, consumer needs sufficient exposure to the stimulus. Opportunity to process: sufficient sensory access to stimuli (size, location, distance, movement, lighting conditions, loudness), pace (speed of movement, rate of speech).

Attention is limited.  Attention depends on level of involvement (greater in context of active purchase decision, lower elsewhere as in post-sale confusion allegations).  Stimuli are more likely to attract attention, all else equal, if they are large or intense, vivid in color, contrast w/background, are centrally or prominently located. Not just driven by physical characteristics; greater for familiar and recognizable stimuli.

Next, perception: the mental registration of sensory inputs into a coherent, unitary whole. Perception is holistic and strongly oriented towards organization and pattern matching, not compositional/feature by feature. We group things that are proximate, we fill in missing info that seems consistent. Subject to least noticeable variation.  Perception is subject to the principle of just-noticeable differences. People can only perceive things that exceed a certain threshold.  Adidas v. Payless: three stripes v. four.  That may not be bigger enough to be recorded.  Five stripes may look different enough.

Perceptions of similarity of A & B are driven by degree of overlap b/t features of A and features of B. Tversky’s theory of similarity. Converse v. Skechers: why do the shoes look similar?  A lot of common features and few not common features.

Categorization: labeling and identifying objects as belonging to a group/category we already know. (Here is an awesome post about categorizing a “dog.”)  Fundamental to human functioning. Identifying a product as belonging to a particular brand is a classic example of categorization; brands function like categories in consumers’ minds. Categorization is often spontaneous, “automatic.” Holistic and based on overall configurations.  Consumers may not be able to verbalize the basis of categorization-based source identifications. Cues used in categorization have three characteristics: they’re observable, typical of the category, and atypical of things that aren’t in the category.

For categorization to take place, the match to typical category cues doesn’t have to be perfect. Everything else equal, stimuli are more likely to be categorized in categories that are highly accessible/well-known brands: a glass of soda will be shorthanded as a “Coke.”

Comprehension: interpreting a stimulus to extract higher-order meaning from it. Inferences based on existing knoweldge. Two common rules: representativeness: attribute to brands that are perceived to be most representative of, or semantically related to the stimulus. Prominence: attribute to brands perceived to be large and prominent in the marketplace. So if I show you a soccer/football championship and ask who’s the sponsor, people will say Gatorade; if I show a chess championship, they guess Microsoft is the sponsor.  Ongoing case: RBX and Reebok shoes—does one suggest the other?

Q: seems to resemble machine learning w/images. Anything to be learned?

Q: dual processing theory? System 1/system 2?

A: don’t like that theory; not a good division. But a little alignment: front end looks a bit like system 1, back end more like system 2.

Q: common for products to have lots of attributes, but consumers are limited in perceiving them.

A: the number of cues you use isn’t not necessarily taxing mentally if you’re pattern matching.  More of an issue: if you have competing cues.

RT: (1) Role of preference for or against cognition? [He thinks it doesn’t mean that they categorize differently in a first pass.] (2) Special problems of trade dress: need some extra principles to figure out what should happen when some of the similarity is due to functional features. [You can control for role of the functional features by changing them in controls.] But that’s not the end of the question: one could have a rule that confusion caused by similarity in unprotectable features must be ignored, or one could have a rule that the defendant is required to stay further away in other ways—add differentiating extras—if it uses those unprotectable features.  [He agrees that part is for lawyers to fight about.]

(3) Sponsorship research on attribution of sponsorship to prominent brands, e.g. Samsung will be more easily perceived as a sponsor of the Olympics just b/c it’s prominent.  But doesn’t that mean that sponsorship, association, affiliation and approval based on similarity as opposed to based on prominence will be difficult or impossible to detect? Is this about what kinds of controls we should be using in sponsorship cases?

Bert Huang, Columbia Law School, Marketing Ethics Through Law

Effects of law on trolley problem: what if we tell subjects (1) criminal law prohibits turning trolley, (2) there’s criminal law, but not enforced, (3) law considers it justified, (4) duty to turn trolley, but not enforced, (5) legal duty to turn trolley.  Can knowing law influence moral intuitions?  People who learn it’s criminal tend not to say it’s morally required to turn the trolley; if it’s required, they tend to say it is.  Asked if it’s morally prohibited to turn, they are more likely to say yes if it’s legally prohibited.s

Looking for: Ethical questions where ethics can go both ways, and law can believably go both ways, and scenario is easily imagined.

For example: search engine paid results. Google Assistant gives oral results; should it have to label ads? Is it ethical?  Ask them what if the law said there’s a need to disclose ads versus it’s fine b/c of the First Amendment.  Or Free app sells your data to marketers, discloses but knows nobody reads the fine print.  What if the law said that wasn’t a real choice v. the law said it’s fine.

Dietary supplements: claims are based on junk science but there are no negative studies showing harm. What if the law said: first, need good science v. claim is fine until harm is proven. [I’d be interested in checking doctor v. marketer as the speaker for this.]

Q: California’s Made in the USA high standard—95% of value has to be made/sourced in the US. New Balance makes shoes that don’t meet that standard: 70% of the value comes from the US.  Expressed in its marketing.  There was a settlement.  New Balance is the only major manufacturer w/manufacturing in the US.

Q: Slack fill: ask if it’s ok. Product packaging: Barbara Kahn’s work on the shape of the package influencing perception of amount.

RT: Wal-Mart’s Equate headache remedy comes in a red box marked migraine and a green box marked headache; the green is much cheaper. The ingredients are the same.  Is that ethical?  Also: Ask them about parol evidence/salesperson who lies!

Q: ask advertisers too what they think is ethical, not just consumers.

Q: why this question? What is the ultimate goal?

A: knowledge.

Q: ethics of pricing: AZT for AIDS when it first came out.

Q: difference b/t political speech and advertising: why is lying ok in #1 and not #2?

Q: do people favor monopolists v. competitive markets?