Social proof is the tendency to trust what other people have already chosen. Reviews, ratings, "most popular" labels, and visible counts all work on the same instinct: uncertainty is resolved by watching others.
On screens it is unusually cheap to deploy, because the data mostly exists already. Surfacing this store's bestsellers, a product's review score, or how many people bought something today requires no new content, only a connection to a system that already knows.
The caveat is that social proof has to be true and specific, or it inverts. Generic claims read as marketing and are discounted automatically; a rating carried without its sample size invites the assumption it is small. "Rated 4.6 by 812 customers" persuades where "Loved by customers" does not, and a fabricated count is both a trust problem and, in most jurisdictions, an advertising offence.
Note also that proof of the wrong crowd repels. A busy-restaurant signal attracts, and "popular with tourists" in a local venue is social proof working against you.