Explainers / MECHANICS
What is the difference between recommended and not recommended reviews on Yelp?
Yelp's automated software sorts every review into recommended or not recommended. Only recommended reviews count toward the star rating and the review count. Of the 22 million reviews contributed in 2025, Yelp says 70% were recommended and 17% were not.
Yelp is the only large review platform that publicly keeps two piles. Open any business page, scroll past the last review, and there is a link to reviews that "are not currently recommended". They are real submissions. They carry stars. They change nothing about the number at the top of the page.
What the software looks at
Yelp describes the system as automated software that evaluates every review using "hundreds of signals of quality, reliability, and user activity on Yelp".11 Yelp Trust and Safety, Recommendation software It does not publish the signals. Its support pages give examples of what tends to land a review in the second pile:22 Yelp Support Center, Why would a review not be recommended?
- the reviewer is "a less active user" and Yelp does not know enough about them
- the review "seems like an unhelpful rant"
- several reviews originate "from the same IP address"
- the review suggests bias, such as one "written by a friend of the business owner"
- the review appears to have been requested by the business
That last item surprises business owners more than any other. Yelp tells businesses not to ask anyone for reviews, including customers, mailing-list subscribers, friends and family, and says its software "tries to identify any reviews that appear to have been requested, and not recommend them".33 Yelp Support Center, Don't Ask for Reviews Google, by contrast, tells businesses they may remind customers to leave reviews and offers a link and QR code for the purpose.44 Google Business Profile Help, Tips to get more reviews The same act, sending customers a review link, is encouraged on one platform and penalized on the other.
The 2025 numbers
Yelp's Trust and Safety Report breaks down the 22 million reviews contributed during 2025:55 Yelp Trust and Safety, 2025 Trust and Safety Report
The removed share rose from the year before; Yelp attributes the increase to improved detection and investigation. It also reports filtering out nearly 500,000 suspected AI-generated reviews in 2025.
Is the filter catching fakes or just being fussy?
An academic study gives the best outside view. Luca and Zervas analyzed restaurant reviews that Yelp's algorithm had filtered and treated them as a proxy for review fraud. They found that roughly 16% of restaurant reviews were filtered, that these reviews "tend to be more extreme (favorable or unfavorable) than other reviews", and that a restaurant was more likely to have suspicious reviews when its reputation was weak, meaning it had few reviews or had recently received bad ones.66 Luca and Zervas, Management Science 62(12), 2016, Fake It Till You Make It: Reputation, Competition, and Yelp Review Fraud Chain restaurants, which depend less on Yelp, were less likely to commit review fraud.
Those patterns are what you would expect if the filter were catching a meaningful amount of real manipulation. They do not show the filter is always right. A genuine first-time reviewer with strong feelings looks, to software, much like a paid one.
The advertiser question
The oldest complaint about Yelp is that buying ads changes which reviews show. Yelp's published position is flat: the software "applies the same objective rules to every business and treats reviews of advertisers and non-advertisers exactly the same", and "no Yelp employee can manually override the software".22 Yelp Support Center, Why would a review not be recommended?
The claim reached the Ninth Circuit in Levitt v. Yelp. Small business owners alleged that Yelp extorted advertising payments by manipulating reviews. In 2014 the court affirmed dismissal, holding that the owners had not stated a claim for extortion because, absent a pre-existing right to be free of the threatened economic harm, "threatening economic harm to induce a person to pay for a legitimate service is not extortion".77 U.S. Court of Appeals for the Ninth Circuit, Levitt v. Yelp! Inc., No. 11-17676 (filed September 2, 2014) Read that carefully. The court ruled on whether the allegations, if true, would amount to extortion under the law. It did not conduct a trial on how the software behaves. The Luca and Zervas study does not test the advertiser claim either.
How to use the second pile
For a reader, the not-recommended section is free extra evidence.
- Compare the two averages by eye. If the hidden reviews are much rosier than the visible ones, the business may be prompting friends or customers. If they are much harsher, look at whether the hidden reviewers have any other activity.
- Look at account depth. A hidden review from an account with one review and no photo is the standard case. A hidden review from a long-time user is unusual and worth reading.
- Check dates. A burst of hidden reviews in one week points to a campaign of some kind, friendly or hostile.
- Do not treat hidden as false. Yelp's own wording is that it does not "know enough about" some reviewers. That is a statement about confidence.
What Yelp does not disclose
The signal list, the weights, the threshold between the two piles, and how often a review switches sides. Yelp says recommended status can change over time as the software learns more about the reviewer and the business, so the split you see today is a snapshot.
Questions readers ask
Do not recommended reviews affect the Yelp star rating?
No. Yelp states that they "don't factor into the business's overall star rating or review count".11 Yelp Trust and Safety, Recommendation software
Can a business owner ask Yelp to recommend a specific review?
Yelp says recommendations are fully automated and that no employee can override the software in either direction.22 Yelp Support Center, Why would a review not be recommended?
Will a not recommended review ever become recommended?
It can. Yelp says the set of recommended reviews changes as the software learns more, for example as a new reviewer becomes more active.11 Yelp Trust and Safety, Recommendation software
Method Based on Yelp's recommendation software page, two Yelp Support Center articles, its 2025 Trust and Safety Report, one peer-reviewed study of filtered reviews and the Ninth Circuit's opinion in Levitt v. Yelp. See how we source and check.
Sources
Every source below was opened and read for this page. Links go to the publisher's own copy.
- Yelp Trust and Safety. Recommendation software
- Yelp Support Center. Why would a review not be recommended?
- Yelp Support Center. Don't Ask for Reviews
- Google Business Profile Help. Tips to get more reviews
- Yelp Trust and Safety. 2025 Trust and Safety Report
- Luca and Zervas, Management Science 62(12), 2016. Fake It Till You Make It: Reputation, Competition, and Yelp Review Fraud
- U.S. Court of Appeals for the Ninth Circuit. Levitt v. Yelp! Inc., No. 11-17676 (filed September 2, 2014)