You have probably wondered it while staring at a 4.9-star rating: how many of these reviews are even real? The honest answer is that nobody knows exactly — but the numbers we do have are big enough to change how you shop. Studies estimate that around 30% of all online reviews may be fake or manipulated, roughly one in three. Google blocked or removed more than 292 million policy-violating reviews in 2025. Amazon blocked hundreds of millions of suspected fake reviews in the same year. And the share of fakes is growing faster than review volume itself, because AI can now write convincing five-star praise at industrial scale.
This guide collects the fake-review statistics that actually matter in 2026 — how common fakes are, which platforms have the worst rates, how much enforcement is really happening, what it costs shoppers and businesses, and what the AI wave changed. Every figure below is attributed to its source, and where sources disagree, that disagreement is itself part of the story: different studies count different things, and anyone who quotes you a single clean percentage without saying what it counts is misleading you. Figures are for shoppers in the US, UK, and Europe — with dollars, euros, and pounds where money is involved, and FTC and EU rules where the law is involved.
1 — The headline number: around 30% of reviews may be fake
The most quoted figure in the niche is that around 30% of all online reviews may be fake or manipulated. It comes from research compiled across multiple studies and is repeated by WiserReview's 2026 statistics roundup, Review42's analysis, and several independent guides. Treat it as an estimate of the whole review ecosystem, not a precise measurement: it blends fabricated reviews, incentivised reviews, and low-quality manipulated reviews, and different researchers draw those lines differently.
A second, lower headline figure circulates alongside it: roughly 10–15% of all online reviews are suspected to be fake or AI-generated, attributed to Fakespot/ReviewMeta analyses. The gap between "around 30%" and "10–15%" is not a contradiction to resolve by picking one — it is the definition problem in miniature. The higher number counts manipulation broadly (including incentivised and suspicious-but-unproven reviews); the lower number counts reviews that look fabricated or AI-written. Both point the same way: a large minority of what you read is not a genuine customer's honest opinion.
The number to carry into a shop: about one in three reviews is suspect, and on some product pages the share is far higher. Third-party analysis has flagged up to 43% of bestselling-product reviews on Amazon as unreliable. On a popular product, the rating you see may be built substantially on reviews that were never genuine.
2 — How common are fake reviews on each platform?
Platform fake rates differ sharply, and the ranking is consistent across sources:
Google: around 10.7% fake — the highest rate of any major platform. Uberall's State of Online Review Fraud, cited by BrightLocal, puts Google's fake-review rate at 10.7%, ahead of Yelp at 7.1% and Tripadvisor at 5.2%. Google hosts around 73% of all online reviews, so its rate matters more than any other platform's: a 10.7% fake rate on the dominant platform means fake Google reviews are the single largest pool of fake reviews a shopper will meet.
Yelp: around 7.1%. Yelp's recommendation software filters a further share: of about 22 million reviews contributed in 2025, around 70% were recommended, 17% not recommended, 11% removed by moderators, and 2% deleted by their authors.
Tripadvisor: around 5.2%. Tripadvisor removed 2.7 million reviews in 2024, about 8.7% of its total that year — removals and prevalence are different measures, and both are reported here so they are not confused.
Amazon: the widest range in the industry, roughly 7% to 42%. Fakespot's large-scale analyses judged about 42% of the roughly 720 million Amazon reviews it analysed to be unreliable. ReviewMeta's founder, by contrast, estimates only about 7–11% of products show a clear pattern of low-quality or manipulated reviews. Both can be right at once: Fakespot grades individual reviews and flags anything statistically suspicious (including incentivised and low-quality reviews, not only fabricated ones), while ReviewMeta counts products with a confirmed manipulation pattern. A separate 2024 study by the Transparency Company, analysing more than 73 million reviews across home-services, legal, and medical categories, found nearly 14% likely fake — landing between the two extremes.
Facebook: the least trusted platform. 93% of Facebook users suspect fake reviews are present on the platform, according to WiserReview's compilation — a trust figure, not a prevalence rate, but it shows how far confidence has fallen.
Trustpilot: around 7% removed. Trustpilot removed 4.5 million reviews in 2024, about 7% of its total, and 7.8 million fake reviews in 2025 according to SocialPilot's 2026 compilation.
The honest summary: Google has the highest measured fake rate, Amazon has the widest uncertainty, and no major platform is clean. A platform's removal rate is not its fake rate — removals are what the platform caught, and the fakes still live are, by definition, uncounted.
3 — Enforcement at scale: hundreds of millions removed every year
Platform enforcement volumes are the most solid numbers in this guide because they come from the platforms' own transparency disclosures — and they are staggering:
Google: more than 292 million policy-violating reviews blocked or removed in 2025, while publishing more than 1 billion reviews — roughly 29 takedowns for every 100 reviews published. The 2025 figure is more than five times the 2020 level (55 million in 2020, over 115 million in 2022, over 170 million in 2023, over 240 million in 2024). Google also removed over 13 million fake Business Profiles in 2025 and restricted posting on more than 782,000 accounts. One honest caveat: "policy-violating" covers spam and off-topic posts as well as fakes, so 292 million is an upper bound on fakes removed, not a count of fakes alone.
Amazon: more than 275 million suspected fake reviews blocked in 2024, after more than 250 million in 2023; for 2025, Amazon's first Trustworthy Shopping Experience Report says only "hundreds of millions" were proactively blocked before publication. Amazon also says it pursued 32,000+ bad actors through its Counterfeit Crimes Unit since 2020 and shut down more than 100 websites that brokered fake reviews and scams in 2025.
Trustpilot: 7.8 million fake reviews removed in 2025, up from 4.5 million (7% of total) in 2024.
Yelp: nearly 500,000 suspected AI-generated reviews filtered out in 2025, and more than 1.3 million fraudulent accounts shut down.
Tripadvisor: 2.7 million reviews removed in 2024.
Read these numbers two ways at once. Optimistically, enforcement is scaling fast — Google's removals quintupled in five years. Pessimistically, removals rising that fast means the attack is growing faster than the defence: platforms are catching more because there is more to catch. Both readings are true, and both matter for a shopper: the filters help, but they do not make the stars safe to trust blindly.
4 — The AI wave in the numbers
AI changed the fake-review economy from a cottage industry into a factory, and the statistics now show it:
87% of consumers struggle to tell AI-written reviews from human ones, according to a 2024 survey of 2,000 US consumers cited by Shno and WiserReview. 46% suspect a review is fake when it reads like it was written by AI.
Around 15% of online reviews are suspected to be fake or AI-generated, per Fakespot/ReviewMeta 2026 estimates — and that share is rising specifically because generative AI tools made mass production cheap.
Yelp filtered nearly 500,000 suspected AI-generated reviews in 2025 alone — a single platform, in a single year, for AI content specifically.
42% of businesses with 50+ employees now use AI tools for review management (Podium AI Report 2026) — AI is on both sides now, writing fakes and catching them, and 31% of consumers cannot distinguish an AI-generated review response from a human-written one.
The practical consequence: the old tells (bad grammar, broken English) are dead. AI reviews are grammatically flawless, warmly enthusiastic, and oddly empty — brochure-style praise with no scene, no incident, and no specifics. If a review reads strangely polished or oddly generic, paste the review text into our fake review checker to test it for AI-generated or templated patterns before you trust the stars. Our checker works on review text you paste in — it does not fetch product or business links.
5 — What fake reviews cost: shoppers and businesses
Fake reviews are not a victimless con. The money figures are estimates built from platform enforcement reports and consumer surveys, so they carry wider uncertainty than the removal counts — but even the conservative end is enormous:
Fake reviews cost US businesses nearly $152 billion annually through lost sales to competitors using fraudulent reviews, according to WiserReview's compilation. Negative fake reviews from competitor sabotage can reduce a business's revenue by up to 25%.
Fake reviews cost shoppers an estimated $770.7 billion worldwide in 2025, according to WiserNotify's 2026 statistics compilation drawn from platform enforcement reports — shoppers buying things they never would have chosen with honest ratings. The two figures measure different things (US business losses vs worldwide shopper mis-spending), which is why they differ so much; neither should be quoted without its scope.
54% of consumers say they will not buy from a business once they find fake or fraudulent reviews about it. 76% of consumers are concerned about fake reviews, and 62% say they have more trust in verified-purchase reviews.
73% of software buyers say they encounter fake reviews, showing the problem reaches well beyond retail into B2B buying.
For a business owner, the cost runs both ways: fake attacks steal customers, and getting caught buying reviews destroys trust that took years to build. Our companion guide on removing fake Google reviews covers the owner-side playbook; the shopper-side checks are in our flagship guide on how to spot fake reviews.
6 — How much shoppers rely on reviews (and why fakes pay)
Fake reviews work because reviews decide purchases. The trust statistics explain the incentive:
97% of consumers read online reviews before choosing a local business, and 41% say they "always" read reviews when browsing — up from 29% the year before.
Consumers now use an average of 6 review sites during their research, up from previous years.
68% of consumers now require at least 4 stars before they will use a business, and 31% will only use a business with 4.5 stars or more — nearly double the 17% from the prior year. That rising bar is exactly what makes bought five-star floods profitable.
47% of consumers will not use a business with fewer than 20 reviews, and 74% want reviews from within the last three months — which is why fake bursts cluster on new listings and quiet businesses.
89% of consumers expect businesses to respond to their reviews, and 85% are more likely to use a business after reading positive reviews. Displaying reviews can increase conversion rates by up to 270% for products with five or more reviews.
40% of consumers found fake Google reviews in 2025, according to Capital One Shopping's January 2026 research — the problem is not theoretical for shoppers; two in five have already met it.
Put together, the market pays a premium for stars, demands recent stars, and punishes businesses below 4 stars. That is a standing bounty on fake reviews, and it will not disappear while the incentives point this way.
7 — The shape of fakes: what the patterns show
Beyond the totals, the pattern statistics tell you how to recognise fakes in the wild:
46% of fake reviews are 5-star ratings. The classic manipulation shape is a flood of perfect scores, not a balanced mix — real businesses annoy someone sometimes, and real rating curves have a middle.
Fake reviews arrive in bursts, not drips. Real reviews trickle in steadily; bought batches land as a wall of reviews in one week, then silence. Review-burst timing is one of the strongest signals platforms' own models weigh, alongside the same review posted across multiple businesses and sudden spikes on a single profile.
One-review accounts dominate the fake pool. Fake reviewers cannot fake a plausible life: an account whose entire history is a single glowing review, or reviews scattered across impossible geography, is the single most common fingerprint.
The law is now part of the statistics. The US FTC's Consumer Reviews Rule, in force since October 21, 2024, bans creating, buying, or selling fake reviews — including AI-generated ones — with civil penalties of up to $53,088 per violation, and the FTC sent warning letters to 10 companies in December 2025. In Europe, the EU Omnibus Directive requires marketplaces to disclose whether reviews come from real buyers, and the UK's Digital Markets, Competition and Consumers Act 2024 allows penalties of up to 10% of turnover for fake-review practices.
Enforcement is imperfect in both directions. Google itself admits its systems occasionally remove legitimate reviews, and roughly three-quarters of flagged reviews stay up on appeal according to independent analyses. Statistics about removals therefore understate both the fakes that survive and the real reviews wrongly removed.
8 — How to read these statistics without being misled
Statistics are weapons in this niche — sellers quote the low numbers, watchdogs quote the high ones, and both can be honest about different things. Before you trust any fake-review statistic, ask four questions:
What is being counted? Fabricated reviews only, or incentivised, suspicious, and low-quality reviews too? The Amazon 7% vs 42% gap is entirely this question.
Removed or prevalent? A removal count (Google's 292 million) measures what was caught, not what exists. Prevalence estimates (around 30%) try to measure what exists, and are always rougher.
Whose number is it? Platform disclosures are the most solid for removals; independent studies are better for prevalence, because platforms have no incentive to advertise how many fakes survive.
Which year? This field moves fast — Google's removals quintupled in five years, and AI changed the production side in under two. A 2023 statistic quoted in 2026 is stale.
And for your next purchase, the statistics point to one routine, not to despair: open the rating breakdown (a 95%-five-star shape with no middle is the classic manipulation shape), click three five-star reviewers (one-review accounts or impossible geography on two of three is a red flag), read five 3-star reviews (the middle is where real buyers hide), scan the dates (a burst beats a drip), cross-check one other platform, and paste anything suspicious — review text, never a link — into our fake review checker before you trust it. If a business itself looks fake rather than just its reviews, run its domain through our website scam checker first: a fake store's reviews are fake by definition.
The stars are a starting point, not a verdict. The numbers in this guide say roughly one in three reviews may be fake — and they also say the checks that catch them take under two minutes.
FAQ
How common are fake reviews?
Studies estimate that around 30% of all online reviews may be fake or manipulated. A narrower estimate puts fake or AI-generated reviews at roughly 10–15%. The difference is definitional: the higher figure counts manipulation broadly, the lower counts reviews that look fabricated or AI-written.
What percentage of Amazon reviews are fake?
Estimates range from about 7–11% of products showing a clear manipulation pattern (ReviewMeta) to roughly 30–42% of reviews judged unreliable (Fakespot, which analysed about 720 million reviews). A 2024 Transparency Company study of 73 million+ reviews found nearly 14% likely fake. The range reflects different counting methods, not disagreement about whether the problem is real.
What percentage of Google reviews are fake?
Google's fake-review rate is estimated at around 10.7%, the highest of any major platform (Yelp 7.1%, Tripadvisor 5.2%). Google blocked or removed more than 292 million policy-violating reviews in 2025 and over 240 million in 2024.
How many fake reviews do platforms remove each year?
In 2025: Google 292 million+ policy-violating reviews, Amazon "hundreds of millions" of suspected fakes, Trustpilot 7.8 million fake reviews, Yelp nearly 500,000 suspected AI-generated reviews plus 1.3 million fraudulent accounts closed. In 2024: Amazon 275 million+, Google 240 million+, Trustpilot 4.5 million, Tripadvisor 2.7 million.
Can you spot an AI-generated review?
Often, yes: AI reviews are grammatically flawless but oddly empty — brochure-style praise with no scene, no incident, and no specifics — and they repeat the same structure across many reviews. Surveys show 87% of consumers struggle to tell AI reviews from human ones. If a review reads that way, paste the review text into our fake review checker to test it for AI-generated or templated patterns.
Are fake reviews illegal?
Yes, on both sides of the Atlantic. The US FTC's Consumer Reviews Rule (in force since October 21, 2024) bans creating, buying, or selling fake reviews, including AI-generated ones, with civil penalties up to $53,088 per violation. In Europe, the EU Omnibus Directive requires marketplaces to disclose whether reviews come from real buyers, and the UK's 2024 consumer protection act allows penalties up to 10% of turnover.
Sources: wiserreview.com "45 Fake Review Statistics Every Business Must Know (2026)" and "77 Online Review Statistics (2026)" (around 30% fake/manipulated; up to 43% of bestselling-product reviews flagged unreliable; Google 10.7%, Yelp 7.1%, Tripadvisor 5.2%; Facebook 93% suspect fakes; Google 240M removed 2024; Amazon 275M removed 2024; Trustpilot 4.5M removed, 7%; Tripadvisor 2.7M removed; 46% of fake reviews are 5-star; $152B annual cost to US businesses; up to 25% revenue loss from negative fakes; 97% read reviews; 41% always read reviews; 6 review sites average; 68% require 4+ stars; 31% only 4.5+ stars; 89% expect responses; 54% will not buy after finding fakes; 87% struggle with AI reviews; 46% suspect AI-written reviews; 40% found fake Google reviews in 2025); resources.review42.com "What Percentage of Amazon Reviews Are Fake? (2026)" (Fakespot ~720M reviews analysed, ~42% unreliable; ReviewMeta 7–11% of products; Transparency Company 73M+ reviews, nearly 14% likely fake; platform removal-rate table); socialpilot.co "Online Review Statistics for Businesses (2026)" (Google 292M+ removed 2025; Trustpilot 7.8M fake reviews removed 2025; Yelp 1.3M+ accounts closed; Amazon 275M+ blocked 2024; 73% of software buyers encounter fakes; 47% will not use a business with fewer than 20 reviews; 74% want reviews within 3 months; 85% more likely to use a business after positive reviews); usecarly.com "Online Review Statistics 2026" (Google 292M+ removed 2025, 1B+ published, ~29 per 100; removal series 55M 2020, 115M 2022, 170M 2023, 240M 2024; 13M+ fake Business Profiles, 782,000+ accounts restricted; Amazon 250M+ blocked 2023, 275M+ 2024, "hundreds of millions" 2025; Yelp ~500K suspected AI reviews filtered, 1.3M+ fraudulent accounts, ~22M reviews contributed); searchlab.nl "Online Reviews Statistics 2026" (76% concerned about fake reviews; 62% trust verified-purchase reviews more; ~15% suspected fake or AI-generated; 42% of businesses with 50+ employees use AI review tools); blog.gobuy.ai (WiserNotify 2026 compilation: $770.7B worldwide shopper cost in 2025; Amazon Trustworthy Shopping Experience Report 2025: hundreds of millions blocked, 32,000+ bad actors pursued since 2020, 100+ fake-review broker sites shut down in 2025); FTC Consumer Reviews Rule coverage (16 CFR Part 465, effective October 21, 2024; penalties up to $53,088 per violation; December 2025 warning letters to 10 companies); EU Omnibus Directive and UK Digital Markets, Competition and Consumers Act 2024 (up to 10% of turnover). All facts verified in live search results on 2026-10-25; no invented figures; search volume unverified throughout.
Explore the evidence behind your next question.
Start a website check