Social Proof

011 min

Social proof at a glance

• What it is: A buyer who cannot judge a product alone treats other people's choices as information about its quality.
• Who defined it: Robert Cialdini, as one of his principles of influence, in 1984.
• What it costs to ignore: In one field experiment, naming a restaurant's five most popular dishes raised demand for those dishes by 13 to 20 percent.
• How to measure it: Split traffic at random and compare conversion on a page with the social evidence against a page without it.
• Watch for: Invented or distorted evidence. Regulators and customers both punish it.

021 min

Why social proof matters

A page that shows ratings, customer counts or logos is making a claim about what other people did. A product manager or a founder owns that claim, and it carries a cost either way. When it is true, it answers a doubt that the buyer cannot settle alone. When it is false, it becomes a legal risk.

The US Federal Trade Commission has made that risk concrete. The Federal Trade Commission announced a final rule that bans the sale or purchase of fake reviews and testimonials and lets the agency seek civil penalties against knowing violators. The rule also covers fake social media followers and views.

A team that treats social proof as copy to fill in, rather than as a statement of fact, has an exposure that grows with every number it shows.

032 min

What social proof is

Social proof is a way of deciding when you lack the information to decide alone. A person facing an unfamiliar choice asks a short question: what are other people doing? The answer works as evidence. Robert Cialdini describes it this way on his own site: "Especially when they are uncertain, people will look to the actions and behaviors of others to determine their own."

Two conditions make it work.

• Uncertainty. A buyer who already knows what they want has little use for other people's choices. A buyer choosing between unfamiliar options has a lot of use for them.
• Similarity. The more the other people resemble the buyer, the more their choices count. The hotel case later in this article shows this.

Products show social proof in several forms:

• star ratings and written reviews
• counts, such as the number of customers or downloads
• logos of known customers
• "most popular" and "best seller" labels
• live signals, such as how many people are viewing an item now

All of them carry the same message: other people chose this. They differ in how specific the message is and in how much the reader resembles those people.

Social proof is not a claim that the product is good. That claim comes from the company, and buyers discount it. Social proof moves the claim to other people, which buyers trust more. It is also not proof in the logical sense. A popular choice can be a poor one, and the buyer is only guessing that the crowd knows something.

041 min

Where social proof comes from

The idea has a laboratory base that is older than the marketing term. The most famous study of social proof is Muzafer Sherif's 1935 experiment. Sherif placed people in a dark room in front of a small light. The light was not moving, but a still light in the dark appears to drift, so every person saw some movement and each person gave a different estimate. Each person was paired with two others and asked to say aloud how far the light was moving. Even though they had given different estimates before, the groups came to a common estimate.

The marketing term comes from Robert Cialdini. Robert Cialdini coined the term social proof in his 1984 book Influence. He presented it as one of the shortcuts that people use when they decide.

051 min

A social proof example: popular dishes on a menu

Hongbin Cai, Yuyu Chen and Hanming Fang tested social proof in a real restaurant.

Before. Customers chose from a menu. For many of them, most dishes were unknown, so a dish had no signal about whether other people liked it.

The change. The researchers varied the information at random. Some customers saw a ranking of the five most popular dishes. The researchers called the learning from other people's choices observational learning. They also wanted to rule out a simpler explanation, called the saliency effect, in which a dish sells more only because the list draws attention to it.

The result. In a randomized field experiment in a restaurant, Hongbin Cai, Yuyu Chen and Hanming Fang found that when customers were given ranking information of the five most popular dishes, demand for those dishes rose by 13 to 20 percent. They tested for the saliency effect and found no significant saliency effect.

The second finding matters for a product team. Showing a list of dish names was not enough. The list of popular dishes carried the effect because it said what other customers had chosen. The authors also report modest evidence that the effect was stronger among infrequent customers, who knew the menu least.

062 min

A second case: hotel towel cards

Cialdini reports a second case that changes one variable: who the other people are. It comes from a field study of hotel guests, in which cards in bathrooms asked guests to reuse their towels.

The first card named the benefits that reuse has for environmental protection. Cialdini says this card led to around 35% compliance. The second card added a social line. Cialdini describes the step this way: "So what would happen if we took a lesson from the Principle of Social Proof and simply included that information on the cards and said that 75% of our guests reuse their towels at some time during their stay, so please do so as well." He reports that towel reuse rises by 26% with this card.

In a third version, the card referred to the people who had stayed in that same room. Changing just a few words on a sign to honestly point out what comparable previous guests have done was the single most effective message, a 33% increase in reuse.

The restaurant case shows that a list of what others chose changes behavior. The hotel case adds that the closer the reader is to the people named, the more the message counts. A general crowd works. A crowd the reader resembles works better.

071 min

How social proof shows up in tech

Four roles meet social proof in different decisions.

• Product manager. You decide which evidence appears at which step. A rating next to a product in a list answers a different doubt than a review count on a checkout page. Look at the step where users leave, and show the evidence that answers the doubt at that step.
• Designer. You decide how much the evidence says. A rating based on three reviews gives the reader much less information than a rating based on three thousand. Show the count beside the average, so the reader can judge how far to trust it.
• Engineer. You decide where the numbers come from. A counter that reads from real events is evidence. A counter that returns a stored or random value is a false statement, and the FTC rule above applies to it.
• Founder. You decide whose names appear. A logo wall is useful only when each customer agreed to appear and is a customer. Choose logos of companies that resemble the visitor, because similarity is what makes the signal count.

Marketplaces and app stores already show ratings and review counts next to every listing. A new product enters a page where the reader compares it with those numbers, so a product with no evidence at all looks unproven.

081 min

How to use social proof

This order follows from Cialdini's two conditions, uncertainty and similarity. It is a working sequence built from them, not a framework that Cialdini published. Each step depends on the one before it.

1. 1

   Find where buyers are uncertain.

   Look at the steps where users leave, at the questions support receives, and at what people search for about your product. Social proof placed where buyers are not uncertain adds nothing.

2. 2

   Find a true fact about other people.

   You need a measured count, rating or share, from your own data. If you have none, stop here. Step 3 cannot start without a fact, and an invented one is the failure the FTC rule targets.

3. 3

   Choose the closest group.

   Among the true facts you have, pick the one about people most like the reader: the same role, the same company size, the same situation. The hotel case shows why.

4. 4

   Place it at the uncertain step.

   The step from the first stage decides the position.

5. 5

   Test it.

   Compare against a version without it, as described below.

091 min

How to measure social proof

Run an A/B test. Split visitors at random into two groups. One group sees the page with the social evidence. The other sees the same page without it. Compare the conversion rate at the step where the evidence appears. Decide the sample size and the end date before you start.

Add a third version when you can: the same item with a label that draws attention but says nothing about other people, such as a plain "featured" tag. The restaurant researchers did this kind of comparison to separate the effect of other people's choices from the effect of attention alone.

Then look past the first purchase. Check returns, refunds and later ratings for each group, because evidence that persuades the wrong buyers can raise sales and lower satisfaction.

101 min

Common mistakes with social proof

Three beliefs lead teams to misuse social proof.

Trap 1: "Any statement about the crowd helps." A statement about average behavior can push people the wrong way. Schultz and colleagues ran a field experiment on household energy use. A message with average neighborhood usage produced either desirable energy savings or the undesirable boomerang effect, depending on whether households were already consuming at a low or high rate. Adding an injunctive message, which conveys social approval or disapproval, eliminated the boomerang effect. If your message names an average, check what it does to people who are already better than the average.

Trap 2: "Numbers that look real are good enough." The FTC rule prohibits anyone from selling or buying fake indicators of social media influence, such as followers or views generated by a bot or hijacked account. The test is whether the number is true, not whether it looks true.

Trap 3: "Any crowd will do." A general crowd works, but the hotel case shows that a crowd the reader resembles works better. A count of users from a different market or a different company size may not carry over.

111 min

Social proof vs. nearby concepts

Social proof is easy to mix up with three neighbours. The deciding question is what the buyer is reading from the signal.

The bystander effect is the same habit of watching others, applied to emergencies. A person who sees others do nothing may conclude that no help is needed. Social proof in marketing is the same habit applied to buying.

121 min

Where the evidence on social proof is contested

The strongest case against trusting social proof comes from two large experiments on how visible choices change what people choose next.

Matthew Salganik, Peter Dodds and Duncan Watts built an artificial music market. It had 14,341 participants who downloaded previously unknown songs, some with and some without knowledge of previous participants' choices. Increasing the strength of social influence increased both inequality and unpredictability of success. Success was also only partly determined by quality: the best songs rarely did poorly, and the worst rarely did well, but any other result was possible.

Lev Muchnik, Sinan Aral and Sean Taylor ran a randomized experiment on a social news site. Positive social influence increased the likelihood of positive ratings by 32% and created accumulating positive herding that increased final ratings by 25% on average. Herding means that each new person follows the earlier ones, so an early lead grows. The authors found that negative influence was corrected by later users, so the effect was uneven.

For a practitioner, this means a display of popularity does not only report quality. It also amplifies early luck. Do not rank a list by popularity alone, and treat early numbers with care. The studies do not say social proof should be removed. They say it moves outcomes away from quality, and you should know by how much.

131 min

How social proof changed since

The meaning of social proof has stayed close to its origin, but the setting has changed. Sherif's 1935 work studied judgments of a light in a room. Cialdini's 1984 book moved the idea into persuasion and selling. Cialdini's own list of such shortcuts now has seven items, and social proof is one of them.

The larger change is that online products turned social proof into data. Platforms record ratings, counts and sales, so effects like those in the restaurant study can now be measured on a live page. The same records made the bias studies possible. The latest development is legal: the FTC rule treats fake reviews and fake follower counts as things a business can be fined for. Current practice has mostly kept Cialdini's meaning, but it now has to be backed by real records.

?7 questions

Questions people ask

§7 sources

Sources