Online Review

Online reviews let anyone tell strangers what they thought of a hotel, a book, a plumber or a phone case. Amazon allowed customer reviews from its early years in the 1990s, and sites such as Yelp and TripAdvisor were built around them. Today they're attached to nearly every listing on the web, usually as a star rating with optional text.

They work because they pool experience. A thousand buyers together know more about a product than any single ad or critic, the idea behind the wisdom of crowds. Shoppers rely on them heavily, and a small difference in average rating can shift sales and bookings noticeably; see Social Proof.

Reviews are also a skewed sample. People who loved or hated something are far more likely to write about it than people who found it fine, so ratings on many sites form a J-shaped distribution with a pile of five-star reviews, a smaller cluster of one-star ones and few in between. The average then says less than it seems to, and the reviews a shopper reads are not a random slice of buyers; see Selection Bias. Negative reviews tend to weigh more heavily on readers than positive ones of equal strength. Ratings across whole platforms also drift upward over time, so a 4.2 may now signal a problem.

Because reviews sell things, they get manipulated:

• Fake reviews are written by sellers, paid services or friends. Large-scale review-for-pay operations have been uncovered repeatedly, and language models make plausible text cheap to produce; see Astroturfing.
• Incentivized reviews offer free products or refunds in exchange for a review, which tends to inflate ratings even without explicit instructions.
• Review bombing floods a product with negative ratings for reasons unrelated to its quality, such as a game studio's political statement or a business owner's viral controversy.
• Suppression includes contracts that forbid customers from posting negative reviews, which the US banned in 2016 for consumer contracts.

Platforms fight back with purchase verification, filters that hide suspicious reviews and lawsuits against fake-review brokers. Detecting fakes is an arms race, since sellers use networks of real-looking accounts to defeat rules based on reviewer identity. In 2024 the US Federal Trade Commission finalized a rule making it illegal to buy or sell fake reviews.

Review scores also become targets. Businesses hound customers for five stars, workers in the gig economy can lose access to a platform over a dip in their rating, and anything less than perfect comes to read as a complaint; see Goodhart's Law.