Celebrities

Yelp Fake Review Alerts Can Cut Restaurant Traffic for More Than a Year

· 5 min read

For restaurants caught planting reviews on Yelp, the penalty may last far longer than the platform’s 90-day warning. A new working paper finds that flagged businesses often remain below their previous foot traffic levels even 18 months later. That’s a meaningful gap between the length of the public notice and the possible length of its business effects. The result puts a concrete cost on a tactic meant to make a restaurant look more popular, though it doesn’t establish that every flagged restaurant will follow the same path.

The warning outlasts its 90 days

Since 2023, Yelp has placed its “Consumer Alert” banner on more than 4,900 businesses accused of posting planted or incentivized reviews. The notice stays on a business’s page for 90 days, giving customers a visible reason to question the reviews while the alert remains. But the research suggests customers may keep their distance after the banner is gone. In other words, removing the warning from the page doesn’t necessarily remove the suspicion it created.

Yi Cao, an assistant professor of accounting at George Mason University and co-author of the paper, says foot traffic at a flagged restaurant fell by 8 percent almost immediately. It dropped further around the holidays, when diners tend to rely more heavily on Yelp. That timing hints at a practical mechanism: if customers consult reviews more often while choosing where to eat, a warning may carry extra weight during those decisions. Across the penalized restaurants studied, many failed to regain their pre-alert traffic even after a year and a half.

The researchers compared Yelp performance with location data from SafeGraph and credit card transactions provided by Consumer Edge. Those sources capture different parts of the same problem: location data can indicate visits, while transactions provide a view of spending. That combination helps show how an online warning translates into visits and spending, though it doesn’t by itself explain every customer’s decision. The pattern points to a lasting trust problem: the public notice is temporary, but the business impact can persist. This is more significant than it looks, because a short-lived alert can affect choices long after customers stop seeing it.

Review manipulation carries a growing detection risk

Yelp’s review ecosystem has changed considerably since the company launched in 2004. Its original email referral service struggled until users began posting unsolicited reviews through a link on the site. In 2005, Yelp gave those reviews a more prominent home, backed by a database of 20 million businesses. The shift helped make customer recommendations a visible alternative to traditional restaurant criticism. It also made the credibility of those recommendations central to the service: when people use reviews to choose a restaurant, they need some reason to believe the accounts reflect real experiences.

That openness also created room for unreliable reviews. Businesses or people paid by them once supplied many of the fakes. Today, the sources can include paid influencers, bot farms and AI-generated text. Each presents a different challenge for a review platform: a paid endorsement may look like ordinary praise, while automated text can imitate the familiar shape of a customer account. A 2020 case involving fast-casual chain Burgerim illustrates how large the problem can become: Yelp placed more than 200 warning banners across its locations after finding fake reviews. The example shows the scale a campaign can reach, but it doesn’t tell us how common that scale is across the industry.

Cao says the research team doesn’t know every signal Yelp uses to spot manipulation. The company relies substantially on reports from consumers and business owners, and runs an algorithm that screens for AI-generated language as a possible sign of review inflation. That mix of reporting and automated screening also makes clear that detection isn’t a simple matter of spotting a single suspicious phrase. As detection improves, a fake review may be more likely to trigger the kind of public warning that drives customers away. And yet, the paper doesn’t show that every warning comes from the same detection process or has an identical effect.

“This is a classical theory of a trade-off between the benefits of manipulating reputation and the risks of getting caught,” Cao says. When the perceived risk is low, he explains, business owners may decide that the potential gains justify it. The study suggests that calculation can miss the long tail of reputational damage. A business might be trying to improve how it appears in search results, but the tactic can put the underlying credibility of its reviews in doubt. The downside may persist after the platform’s formal penalty ends.

Authenticity is the harder asset to replace

Yelp reviews aren’t permanent verdicts, and a restaurant’s reputation can change with its service and food. That matters because the findings describe a lingering effect, not an irreversible sentence for a business. Still, the results make review manipulation look like a costly shortcut: the alert lasts three months, while the traffic decline may endure far longer. The distinction is useful for reading the research carefully; a prolonged drop in visits signals a serious problem, but it doesn’t mean recovery is impossible.

As AI-generated reviews grow more polished and harder to distinguish from real customer accounts, diners have more reason to value signs of genuine experience. The paper’s results suggest that a platform warning can change behavior beyond the period when the warning is displayed. (And this is the part most people overlook.) Trust isn’t restored simply by waiting for a banner to disappear. Hard to earn back.

If you’re working in this space, the findings are a reminder that review systems depend on more than the number of stars on a page. Restaurants need customers to believe that the feedback reflects actual visits, and platforms need ways to make manipulation costly enough to deter it. The research doesn’t settle how much of the traffic decline comes from the banner itself, continued customer suspicion, or other changes at the businesses. But it does show why a temporary public notice can have consequences that last well beyond its stated run.

For restaurants, the practical takeaway is clear: a credible record of customer feedback is harder to rebuild than a rating is to inflate. That’s not a guarantee that every honest business will recover quickly, or that every suspicious review will be caught. It is a reason to treat authenticity as a business asset rather than a cosmetic detail. The numbers here are underwhelming for anyone hoping a short alert will produce only a short setback.

Source: Francky Knapp · www.eater.com