When in doubt, do what others do.
When people aren’t sure what to do, they copy what others are doing and assume it’s right. Reviews, ratings, and usage numbers put that instinct to work, as long as the proof is real and the norm is worth following.
Robert Cialdini coined the term in his 1984 book Influence, where it is one of six principles of persuasion, alongside reciprocity, commitment and consistency, authority, liking, and scarcity. In uncertain situations, people assume that others know more than they do, and follow them.
The best-known early evidence comes from Muzafer Sherif’s 1935 experiment. In a dark room, a still point of light seems to move, and each person judged the motion differently. In groups, their estimates converged on a shared answer, and people kept the group’s answer when tested alone again.
When people are unsure how to act, they look at what others do and assume it is correct. Social proof is strongest when the situation is ambiguous, when being right matters, and when the others seem similar or knowledgeable.
In Sherif’s experiment, a still light in a dark room seemed to move by different amounts for different people. Together, their estimates converged, and each person kept the group’s answer afterward. With no clear answer, the group became the answer.
Testimonials and ratings from earlier customers encourage new ones to sign up and buy. Show the distribution of ratings and recent reviews, not just an average, so people can judge for themselves.
Social proof is strongest from people we see as similar. In a door-to-door charity drive, a longer list of earlier donors brought more donations, especially when the names were friends and neighbors. Show testimonials from the visitor’s role, industry, or region.
Hotel guests told that most guests reused their towels reused theirs more often than guests given an environmental appeal, and more still when the message referred to guests who had stayed in their own room.
When households were told the neighborhood’s average energy use, those below average started using more. Adding an approving smiley face for low use prevented it. Never advertise a low norm, such as how few people have finished their profile.
In the MusicLab experiment, showing download counts made hit songs more unequal and less predictable: early popularity snowballed. Popularity sorts make the popular more popular, so also offer sorts such as newest or best match.
Follower, like, and view counts shape how trustworthy an account seems, and the same goes for used by 10,000 teams on a landing page. Keep such numbers current and honest, and show them only when they’re impressive.
Planted applause and laugh tracks are old tricks; fake reviews are the online version. In 2024 the US Federal Trade Commission finalized a rule banning fake reviews and testimonials, including ones generated by AI.
AI can generate convincing reviews in seconds, and it learns from what is already popular. Both make honest social proof harder and more valuable.
AI makes fake reviews cheap and convincing, and it also summarizes real ones. Label AI summaries, link them to the original reviews, and verify that reviewers actually bought or used the product.
AI recommendations echo what is already popular in their data, a social proof loop at scale. Surface relevant but less popular options too, and explain why something is recommended.