Home › Explainers

Why Google Reviews Disappear: The Seven-Day Filter

Published October 31, 2026

A customer swears they left your business a five-star review. It never appears — or it appears, and a week later it is gone, with no notice and no explanation. In 2026, a research team tracked 4.9 million Google reviews across 9,048 business profiles for two weeks and counted what vanished. Their finding reframes the mystery: disappearance is not an occasional glitch. It is a standing filter that removes roughly 2% of new reviews in their first week, every week — and it follows rules you can learn.

First, separate the bugs from the filter

Two different things make reviews vanish, and they need opposite responses. The first is a platform bug. In February 2026, reviews disappeared from business profiles while remaining in the profile's review count and still visible on the reviewers' own profiles. Google acknowledged a display issue on February 10, called it mostly fixed on February 13, and by mid-March most reviews were back. When the count and the visible list disagree, that is the signature of a display problem: wait.

The second is deliberate removal. With no incident running, the researchers watched 4,590 reviews disappear from their panel in two weeks in August 2026. That is the filter: Google's automated screening, which the company says blocked or removed over 292 million policy-violating reviews in 2025 while publishing more than a billion. If a review is gone from the reviewer's own profile too, it was removed, and it will not return on its own.

A new review is on probation for seven days

Every review passes a check before it is published, which Google says takes a few days at most. The study's central finding is what happens next: deletion risk follows the review's age, not the calendar. Negative reviews are screened first, peaking on days two and three. Positive reviews are screened later — 47% of the positive reviews removed in their first 13 days disappeared on days six and seven. After day eight, the daily risk falls to about a tenth of the day-seven peak and stays there.

Old reviews are not exempt. Half the deletions in the panel hit reviews under a month old, but a third hit reviews over a year old — most of those in bulk, when Google removed the author's entire account. One deletion in five, where the author could be traced, was every review by that account disappearing on the same day, across unrelated businesses.

The account matters more than the text

The single strongest predictor of removal was not anything the review said. A first-ever review was deleted 5.6 times more often than a review from a Local Guide at level 6 or above. First reviews made up 14% of the reviews in the panel and 22% of the deletions. The same words, from a new account, are treated as a different object than the same words from an established one — Google's models weigh the account, including its links to other suspicious accounts, far more than the sentence.

This has an uncomfortable consequence for honest businesses: the customers least likely to have an established Google presence — older customers, one-time visitors — write the most fragile reviews. It also explains a pattern our Verified Purchase explainer found on Amazon: platform trust signals certify the account and the transaction, not the sincerity of the text.

Extremes go first, and bursts look suspicious

Within the same business, one-star reviews were deleted 3.9 times more often than four-star reviews. About 5.7% of new one-star reviews were gone within a week, against 1.8% of five-star reviews and under 1% of three- and four-star reviews. Google removes the extremes first, and one-stars fastest of all.

Timing matters as much as tone. A review posted on a day when the business received at least three positive reviews and four times its usual daily volume was deleted 1.8 times more often — the risk rises with the day's count, while a high-volume month carries no extra risk at all. The filter reacts to the burst, not the trend. Perfection is also a signal: profiles where 85% or more of reviews are five stars lost reviews at 2.2 times the rate of profiles in the 50–70% range, and on profiles rated 4.8 or above, even a four-star review was deleted nine times more often than on profiles under 4.5. That profile-shape risk sits alongside the solicitation rules in our Google 2026 review rules explainer: how you collect shapes what survives collection.

What makes no difference at all

The study is equally useful for what it cleared. Reviews naming a staff member were deleted at exactly the same rate as reviews that did not (1.01×) — despite Google's April 2026 policy against soliciting reviews that identify staff, the filter does not appear to enforce it. Review length made no difference. Stock phrases such as "highly recommend" made no difference. Photos looked protective in the raw numbers, and the effect vanished once other factors were controlled. If you are editing your review requests to dodge the filter, most of the usual folk advice is aimed at things the filter ignores.

The same filter removes review bombs

Before treating every deletion as a loss, note the direction of the totals. One- and two-star reviews are 9% of all reviews but 20% of deletions. When two Spanish gyms in the panel received 297 one-star reviews across August 1–2 in an apparent review bomb, 295 of them were gone by August 16 — the filter cleaned the attack up largely on its own. And because Google removes more one-stars than five-stars, the net effect on scores was tiny: among businesses that lost five or more reviews, the average rating moved by 0.01 of a star. Deletion feels personal; arithmetically, it mostly cancels out.

What actually helps

• Reply to positive reviews within two days. Among new five-star reviews, 1.5% of those with an owner reply by day two were deleted between days three and seven, against 2.4% without one — at equal sector and size, the odds were roughly halved. Any reply by day two counts; there is no same-day race. It did nothing measurable for one-star reviews.
• Spread requests across days. Send the invitation when the job closes or the order lands, not to the whole customer list on one Monday. A busy month is fine; a busy day is the risk pattern, and a burst repeated weekly is the pattern behind posting restrictions described in case work by local-search practitioners.
• Keep your own copy. Google never tells you which review went. Export or screenshot your reviews monthly, so a disappearance is something you can identify and, for a genuine review removed in error, raise with support. Google's own help pages concede the automated system can remove legitimate reviews by mistake — but reviews removed for policy violations are not restored, so only appeal removals you can stand behind.
• Count on day eight. Judge a campaign by what is still live after the probation week, not by the day-one total. That is the same discipline as reading a score by its review count and mix, which our guide to how many reviews make a rating trustworthy and our invited vs organic explainer both argue from different directions.

Sources

• Reviewflowz × Cartelis (2026), "Gone in 7 days" — daily tracking of 4.9 million reviews on 9,048 Google profiles over two weeks in August 2026: ~2% of new reviews removed in week one; day-by-age deletion pattern; 5.6× first-review risk; 5.7% one-star vs 1.8% five-star first-week removal; 1.8× burst-day risk; 2.2× too-perfect-profile risk; 1.01× staff-name effect; 295 of 297 review-bomb reviews removed; 0.01-star average rating movement; reply-by-day-2 figures
• Google (as reported in the study above): over 292 million policy-violating reviews blocked or removed in 2025, more than a billion published, posting restrictions on 782,000 accounts; February 2026 display issue acknowledged February 10 and described as mostly fixed February 13, 2026.

Data checked October 31, 2026. The deletion figures come from one research panel of Google business profiles in August 2026, published by a review-management company; they describe Google's behaviour as observed, not Google's internal rules, and rates vary by country and sector (US profiles in the panel lost about twice as much as European ones). Nothing here is a guarantee that any individual review will stay or go.

Summary

Google removes reviews constantly and quietly: in the 2026 Reviewflowz × Cartelis panel, about 2% of new reviews vanished within a week, screened by age (negatives on days 2–3, positives on days 6–7), with first-time reviewers deleted 5.6 times more often than experienced Local Guides and one-star reviews removed fastest of all. Bursts and near-perfect profiles attract extra risk; wording, length and staff names do not matter. The same filter removes most review bombs by itself, and ratings barely move (0.01 stars on average), while replying to positive reviews within two days roughly halved their deletion odds.

The single most important takeaway: a review is not yours until it survives day eight — so collect steadily instead of in bursts, reply early, and judge every campaign by what is still live after the probation week.

← Back to the rating wall