Social proof

In short

Picture two restaurants in a town. One is empty; in the other, every table is taken. If you don’t know the food, you very likely go to the full one. The name for that decision is social proof.

The same thing happens on screen. The number of reviews, the average rating, a “Bestseller” label and customer logos tell the person what others do. With that information, they reduce their own uncertainty.

The problem: this information is easy to make up. Below I first describe the honest use, then the fake forms.

Origin of the term: Cialdini

The term was coined by psychologist Robert B. Cialdini. The book appeared in 1984 as Influence: The Psychology of Persuasion. The 1985 edition is titled Influence: Science and Practice.

In the book, Cialdini lists principles of persuasion, and social proof is one of them. The others are reciprocity, commitment and consistency, liking, authority and scarcity. Unity was added as a seventh principle in the most recent edition.

According to the Wikipedia article, the effect gets stronger under two conditions. The first is uncertainty: when people don’t know what to do, they look to others more. The second is similarity: people adopt the behaviour of those who resemble them more easily.

Cialdini, Noah Goldstein and Vladas Griskevicius tested this in two field experiments in hotels. The study appeared in 2008.

A sign saying “the majority of guests reuse their towels” worked better than a sign that appealed only to environmental protection. Most effective was the message describing the behaviour of guests in the same room.

The similarity condition is a useful hint for interfaces. The sentence “Small businesses like yours choose this plan” can do more than a general “thousands of people”. Provided it is true.

Live example: the same product card, three states

The card below comes from a fictional kitchen store. The product and price are the same in every state. The only thing that changes is what the card tells you about others.

The second and third cards have the same goal: more sales. The difference is whether you can check the information. The honest card shows the number and the source; the manipulative one only creates a feeling of urgency.

Honest use in the interface

One of the studies measuring the effect of reviews on sales comes from the Spiegel Research Center at Northwestern University. The center’s 2017 report is based on data from three online stores. The center analysed this data together with a review software provider.

• At the store selling high-priced gifts, the purchase likelihood of a product with five reviews was 270% higher than for a product with no reviews.
• At the same store, reviews increased conversion by 190% for a low-priced product and by 380% for a high-priced one. Risk rises with price, and reviews reduce that risk.
• Purchase likelihood usually peaked at ratings between 4.0 and 4.7 and fell as the rating approached 5.0. The report explains this with a “too good to be true” impression.
• Seeing a review from a verified buyer increased purchase likelihood by 15% compared with an anonymous review.

These figures come from the data of three stores and don’t carry over one-to-one to every product. Still, the direction is clear: real, checkable reviews help undecided people.

On product pages, I stick to these rules:

• Put the number of reviews next to the average rating.
• Date reviews and mark verified purchases.
• Don’t hide negative reviews; make them findable with a filter.

Details of this layout are in the product page pattern.

Be careful with small numbers. In Jen Cardello’s article at Nielsen Norman Group, a participant found an article shared by 1,000 people on Facebook not popular enough.

According to Cardello, the biggest risk of social proof is the perception that too few people approve of the content. The article also says that social elements can clutter the interface and slow down the page.

Social proof isn’t limited to e-commerce. Real customer logos on a pricing page or the real number of users on a SaaS sign-up screen do the same job. The rule is the same everywhere: the number must be measured, and there must be a source behind the claim.

Dark use: the fake “12 people are looking at this”

Because social proof works so well, it is often made up. The study by Arunesh Mathur and his team, published in 2019, crawled about 53,000 product pages from about 11,000 shopping websites. The team found 1,818 dark pattern instances in total.

313 of them were notifications about other users’ activity, such as “35 people have added this product to their cart”. The researchers checked these notifications against the page source. They ignored notifications generated server-side, because they couldn’t verify them.

Among the rest, they found 29 deceptive notifications on 20 websites. Most of these websites generated the notifications at random; even the number of people “currently viewing” came from a random number generator. Others had hard-coded previously generated notifications into the page.

Countdown timers look similar. Of 393 timers the team found, 157 were deceptive, spread across 140 websites. With them, the timer reset after expiring, or the offer stayed valid after it expired.

The team also found 12 customer testimonials of uncertain origin. In one example, the same testimonials appeared on another website under different customer names. Some third-party providers even openly advertised that they could create fake orders.

Designs like these are called dark patterns. I describe their types in the guide to dark patterns. Whether a particular design is legally permitted, clarify with a specialist; this note isn’t legal advice.

How to check it on your own website

You don’t need a special tool to find fake social proof. Mathur and his team examined the timers this way too: they visited the websites repeatedly and saved screenshots of the timer.

1. Reload the product page three times. If the number of people “currently viewing” changes every time, ask your team where it comes from.
2. Wait for the countdown to finish. If the deal doesn’t end at zero, the timer is lying.
3. Put a sentence from a review in quotation marks and search for it. If the same sentence shows up on other websites, the review very likely doesn’t come from your customers.
4. Ask for the source of round claims such as “thousands of happy customers”. Remove numbers without a source from the page.

If you want to know how an honest element affects sales, don’t guess; measure. Calculate the sample size beforehand with the A/B test calculator.

My note

In an audit, I look at social proof with two questions. Where does this number come from? Can the person check it? If I can’t answer yes to both, I write the element up as a finding.

The problem I meet most often is outdated proof. The product page shows two reviews that are three years old; the home page shows the logo of a client you no longer work with. Users notice and think the website is neglected.

“A fake counter increases conversion” is a defence I often hear. The short-term increase can be measured, but the loss of trust in a person who sees the number change is hard to measure in the same test. I don’t recommend that risk to any of my clients.

If you want someone from outside to look at the reviews, logos and counters on your product page, the UX Audit covers this check too.

Related pages

Social proof has the most effect on the product page and at checkout. I describe these two screens in full in the e-commerce UX guide and the checkout pattern.

Before measuring the effect of social proof, clarify how you define conversion. See the guide to the conversion rate. The other principles are in the list of laws of UX.