The Three Biases That Slip Through Most Market Research.

Three of the most reliable influences in behavioral science are also the easiest to miss in research. Here's how to design studies that catch them.

Ask a shopper why they chose one brand over another and you will get a reasonable answer. Better taste. Better price. Better ingredients.

You will rarely hear, "Because a lot of other people seemed to be buying it." Or, "Because there were only two left." Or, "Because a doctor said so."

That isn't because those forces don't matter. It's because people mostly can't see them working. Social proof, scarcity, and authority are three of the most studied influences in behavioral science. Most marketers know the names. Far fewer insights teams build studies that can actually detect them. That's a problem, because these are exactly the drivers that stated- preference research tends to miss. As I've written before, the gap between what people say and what they do isn't a nuisance. It's the core problem.

I spent years running shopper marketing and a retail testing lab at Frito-Lay, and the lesson I relearned most often was simple: what shoppers told us and what they did at the shelf were two different data sets. These three biases live in the space between them.

Social Proof: The Influence People Swear Doesn't Affect Them

Social proof is our tendency to look at what others are doing when we aren't sure what to do ourselves. It's a shortcut, and usually a sensible one.

The research problem is that people consistently underestimate it. In a field experiment by Nolan, Schultz, Cialdini, Goldstein, and Griskevicius, California households received doorhangers encouraging them to save energy. Some messages appealed to protecting the environment, some to social responsibility, some to saving money. One simply told residents that most of their neighbors were already conserving. The neighbor message produced the biggest change in behavior. It was also the message residents rated as the least motivating. A companion survey of 810 Californians found the same pattern: beliefs about what neighbors were doing predicted conservation better than any other belief, even though respondents rated them as the least important reason.

Read that as a researcher. If you had run only the survey, you would have killed the winning
message.

The well-known hotel towel studies point in the same direction, with an important caveat. Goldstein, Cialdini, and Griskevicius found that telling guests most other guests reused their towels lifted reuse to 44.1%, versus 35.1% for the standard environmental appeal, and that a "guests who stayed in this room" version did better still. But when Bohner and Schlüter repeated the work in two German hotels, the norm messages were no more effective than the standard message. Social proof is real. How much it moves people depends on context, culture, and how relevant the "others" feel.

It's also worth being precise about the numbers this field likes to quote. Nielsen's 2012 Global Trust in Advertising survey of more than 28,000 online respondents in 56 countries found that 92% trusted recommendations from people they know, and 70% trusted consumer opinions posted online. Useful context, but notice what it is: a measure of stated trust. It tells you what people say they believe, not what changes their behavior.

What this means for research:

• Don't ask respondents whether social proof influences them. Most will say it doesn't. Test it
  instead, with monadic or A/B designs where the only difference is the presence, source, and
  specificity of the social cue.
• Vary who the "others" are. "Most shoppers" and "people in your neighborhood" are
  different stimuli and can produce different results.
• Measure behavior or a close behavioral proxy. Choice tasks, simulated shelves, and in-
  market tests will tell you more here than a purchase-intent scale.

Scarcity: It's Not Just How Much Is Left. It's Why.

Scarcity is the tendency to value something more when it seems less available. The most cited demonstration is the 1975 cookie study by Worchel, Lee, and Adewole. Participants rated chocolate chip cookies from a jar holding either 10 cookies or 2. The cookies were identical. The ones from the 2-cookie jar were rated as more desirable.

The part that gets left out of most retellings is the more useful part. Each participant saw a single jar, not two jars side by side. And the researchers varied how the scarcity came about. Cookies whose supply dropped from 10 to 2 were rated as more valuable than cookies that had been scarce all along. And cookies that became scarce because other participants had eaten them were rated higher than cookies that became scarce by accident.

In other words, scarcity works hardest when it is a change, and when it signals that other people want the thing. Scarcity and social proof are often doing the work together.

For researchers:

• Specify the scarcity story in your stimulus. "Limited edition," "selling fast," and "short supply
  due to a production issue" are different claims. Test them as different cells.
• Watch for order and context effects. Scarcity is relative, so what respondents saw before, in
  the study or in the market, changes how they read it.
• Be cautious with stated intent. Scarcity acts in the moment of choice, and a survey
  completed at leisure is a different moment.

When you advise brand teams, be clear about the line. The FTC's staff report Bringing Dark Patterns to Light specifically calls out false low-stock messages, false high-demand claims, and
countdown timers that reset as deceptive design. If the scarcity isn't real, it isn't an insight. It's a liability.

Authority: Expertise Changes What People Remember

Authority bias is the tendency to give more weight to information from people we see as experts or legitimate authorities.

The classic demonstration is Milgram's 1963 obedience study, in which 26 of 40 participants continued to the maximum 450-volt level on the instructions of an experimenter. It's a landmark, and a study that could not be run in its original form today. It's also about obedience to a person in the room, which is a long way from a dermatologist's name on a carton.

For consumer work, a more relevant line of evidence comes from Klucharev, Smidts, and Fernández. Using fMRI, they found that a single exposure pairing a product with an expert produced a long-lasting positive effect on both memory for the product and attitude toward it.

Expertise didn't just make a message more persuasive in the moment. It changed what people encoded.

Two cautions matter for research design. First, authority is granted, not declared. "Clinically proven," a professional endorser, and a brand's heritage story are all authority cues, and respondents will grant or withhold credibility differently by category and culture. Second, authority and social proof overlap. "Recommended by dentists" is expert endorsement. "4.7 stars from 12,000 reviews" is crowd endorsement. They are different mechanisms and should be tested separately before anyone combines them.

For researchers:

• Test authority cues as claims, in context, on pack or on screen, against a no-endorser
  control.
• Measure source credibility separately from claim appeal, so you know which one is doing
  the work.
• Build in delayed recall where you can. If part of authority's effect shows up in memory, an
  immediate preference measure will understate it.

The 2026 Wrinkle: Machines Read These Cues Too

AI shopping assistants increasingly sit between people and products. That doesn't make these biases go away. Ratings, review counts, expert badges, and stock messages are exactly the kind of structured signals an assistant can read, weigh, and summarize before a person ever sees the page. The cues still matter. They may simply reach the shopper secondhand.

That raises the stakes on authenticity. The FTC's rule on consumer reviews and testimonials, in effect since October 2024, prohibits fake reviews, including AI-generated ones, and bans buying or selling fake social media indicators such as bot followers. Manufactured social proof is now a legal problem, not just an ethical one.

For insights teams, the question is no longer only "Does this cue move people?" It's also "Is the cue true, and how does it land when a machine reads it first?"

What Researchers Owe the Brand Team

None of these biases is a trick. They are shortcuts people use because most of the time they work. Other people's choices, genuine scarcity, and real expertise are useful information. That's why the research job here is so specific. Our role isn't to help a brand team decide how hard to pull these levers. It's to tell them which cues are true, which ones actually change behavior in their category, and which ones respondents will never mention even when they matter most.

People can't tell you how social proof, scarcity, and authority affect them.

Good research doesn't need them to.