A 2026 Guide for Businesses

Most guides for how to spot fake reviews are written for shoppers deciding whether to buy a blender. This one is written for the person whose business profile is the target.

The signals below apply across Google, Trustpilot, the BBB, industry platforms, and employer review sites. They work whether someone planted a one-star attack on your listing or bought a wall of five-star reviews for a competitor. And because a growing share of fabricated reviews are now written by language models, the second half of this guide covers the signals that still hold when the text itself looks flawless.

Key Takeaways

• A fake review is any rating or comment that does not reflect a real customer experience, whether it attacks your business or inflates someone else’s.
• No single signal proves a review is fake. A defensible case comes from stacking several signals, especially timing, reviewer history, and factual errors about your operation.
• Research on more than 714,000 reviews found AI-generated fake reviews are cleaner and less exaggerated than genuine reviews, which means poor grammar and over-the-top praise are now the least reliable tells.
• A “Verified Purchase” badge confirms a transaction happened. It does not confirm a human wrote the review.
• Reports of fake negative reviews supported by documented evidence get treated differently than reports that simply say a review is unfair.

What Counts as a Fake Review

A fake review is any rating or comment that does not reflect a real customer experience with the business being reviewed. That definition is broader than most business owners expect, and it covers several things businesses run into regularly:

• Competitor-planted negatives: one-star reviews from accounts that never transacted with you, often timed to a promotion, a launch, or a busy season.
• Purchased positives: fake positive reviews bought in bulk by a competitor, to lift their rating, usually from broker networks or social media groups.
• Insider reviews: feedback from employees, owners, family, or vendors posted without disclosing the relationship.
• Incentivized reviews: ratings exchanged for a discount, gift card, refund, or free product, particularly when the incentive is conditioned on leaving a positive review.
• Extortion reviews: negative ratings posted, or threatened, to pressure a business into paying.
• Filtered feedback: systems that route unhappy customers to a private form while sending happy ones to a public platform.

Several of these are not only against platform policy. In the United States, the Federal Trade Commission’s Rule on the Use of Consumer Reviews and Testimonials (16 CFR Part 465), effective October 21, 2024, makes creating, buying, or selling fake reviews unlawful, with civil penalties for review fraud reaching $53,088 per violation as of 2026.

The filtering practice in the last bullet above has its own name and its own enforcement history, explained in our guide to review gating and why platforms prohibit it.

Seven Signals That a Review Is Not Real

No single signal proves anything. Real customers write short reviews, forget details, and sometimes get facts wrong. What builds a case is several signals appearing together on the same review or across the same cluster.

• Timing that does not match your transaction volume: Four reviews in an afternoon at a business that averages four a month is the strongest single indicator available to you, because it is the one signal an outsider cannot see and you can.
• Reviewer history that does not fit your service area: Open the reviewer’s public profile and look at what else they have reviewed. A pattern of businesses spread across states you do not operate in, posted within a short window, points to an account created to post rather than a customer who happened to visit those areas.
• Missing operational detail: Genuine reviews tend to name something specific: a staff member, a service, a wait time, a product failure. A review that could apply to any business in your category, with nothing only a customer would know, deserves a closer look.
• Details that are wrong: This is the most useful signal for a report, because it is verifiable. Reviews that name services you do not offer, locations you do not serve, staff who do not work for you, or products you do not sell were written by someone without firsthand experience.
• Repeated phrasing across reviews: Reviews produced in bulk reuse sentence structures and stock phrases. Reading several suspect reviews side by side quickly reveals this pattern, and screenshots of the overlap make strong supporting evidence of fake negative reviews.
• Language pointing to payment or incentive: Mentions of a gift card, voucher, discount, free product, or refund in exchange for feedback indicate a review that was compensated, which most platforms prohibit and which the FTC rule addresses directly.
• Profiles with no history: A single review, a default avatar, and a generic display name are common to fabricated accounts. Treat this as supporting evidence only, since plenty of real customers review one business and never return.

Once you have flagged a review, cross-reference it against your own records. Search the reviewer name, email, and phone number against your CRM, point of sale, booking system, or ticket history for the date range in question.

A documented absence from your customer records is one of the more persuasive things you can include in a report, and this kind of cross-checking is a standard part of ongoing business review management.

Source: Google Maps 2025 Trust and Safety Report, April 2026.

Why That Checklist Fails on AI-Written Reviews

The checklist above for how to spot fake reviews was built for fakes written by people, and it fails on fakes written by machines because AI-generated reviews do not carry the flaws the checklist looks for.

A study published in the Journal of Retailing and Consumer Services in 2025 analyzed 714,016 reviews across AI-generated fake reviews, human-written fakes, and authentic reviews. It found AI-generated fake reviews were more readable and showed lower levels of specificity, exaggeration, and carelessness than both human-written fakes and genuine reviews.

In practical terms, the polished and grammatically clean review now deserves more scrutiny than the sloppy one, not less. Typos, all-caps, and stacked exclamation points remain useful, but they only catch the least sophisticated attempts.

Verification badges are no safer. In a 2025 analysis, Pangram Labs scanned 30,000 front-page Amazon reviews across 500 best-selling products and identified 909 of them, or 3%, as AI-generated with high confidence. Of those, 74% carried a five-star rating compared with 59% of legitimate human reviews, and 93% displayed the “Verified Purchase” badge.

Verification confirms that a transaction occurred. It says nothing about who or what wrote the words attached to it.

What still works are the signals that never depended on the text. Timing clusters, reviewer geography, posting velocity, factual errors about your operation, and absence from your customer records are behavioral and metadata signals. They hold regardless of authorship, and they are what your report should lead with.

That also explains the limits of fake review checker tools and AI detectors. They score text, so they inherit exactly the blind spot described above. A checker cannot see your appointment book, your service area, or how fast an account has been posting. Use a score from a fake review checker tool as one input among several, never as a verdict, and do not submit it as evidence in a report to a platform.

Telling a Coordinated Attack From a Bad Week

The difference between an attack and a genuinely bad stretch is pattern, not content. Four questions separate them.

1. How does the rate compare to your baseline? Measure review velocity against the same period last quarter and against your actual transaction count for the window. Volume that outruns the number of customers you served is the clearest tell.
2. What shape is the ratings distribution? A real service problem produces a spread, including twos and threes and the occasional four. A coordinated push clusters hard at one end.
3. Does it show up on more than one platform? Genuine service failures surface across Google, industry platforms, and social channels within a similar window. Coordinated activity usually concentrates on a single profile.
4. Do the reviews reference an incident rather than a transaction? Reviews responding to a news story, a social post, or a controversy, without describing a purchase or a visit, indicate a campaign rather than customer feedback.

Platform behavior is itself a signal. Google now responds to detected spam spikes by removing the content, pausing new reviews on the profile, alerting the owner, and displaying a public banner explaining why contributions are paused.

If that banner appears on your listing, the platform has already reached a conclusion about the activity, and you should be capturing evidence rather than waiting.

Catching this early depends on watching the profile continuously, which is the case for ongoing reputation monitoring. When an attack is tied to press coverage or a public incident, it stops being a review problem and becomes a crisis management problem.

What Platforms Catch Before You Do

Most fabricated reviews never reach your profile. In its 2025 Trust and Safety Report, published in April 2026, Google reported blocking or removing more than 292 million policy-violating reviews, restricting more than 782,000 accounts, and removing over 13 million fake Business Profiles in a single year.

That scale has an uncomfortable consequence. If automated screening catches the obvious attempts at that volume, the fakes that survive to reach your listing are the ones built to pass it. They read normally, they arrive at plausible intervals, and they often sit on accounts with some history. This is why “it just looks fake” is rarely enough to get a review removed, and why the documentation step below matters more than it used to.

Standards also differ by platform, and a report that works on one may not translate to others. Our platform-specific guides cover flagging fraudulent Trustpilot reviews and identifying fake Google reviews on a Business Profile.

Document the Review Before You Report It

Report outcomes are based on the quality of the evidence attached to them. A submission that cites a specific policy violation and supports it with records is handled differently than one that simply argues the review is unfair.

What to Capture

1. A full-page screenshot showing the review text, star rating, posting date, and page URL together in one frame
2. The reviewer’s public profile: total review count, other businesses reviewed, locations, and posting dates
3. The result of your customer record search, including the systems checked and the date range covered
4. Every factual error in the review, listed specifically: services not offered, staff not employed, locations not operated
5. A log of the full cluster if more than one review is involved, with timestamps, in a single file
6. The exact policy clause the review violates, copied from the platform’s own published policy

Where to Report It

File through the platform’s own reporting flow and lead with the policy violation, not a description of why the review is untrue. Platforms enforce their policies, not fairness. If the report is declined, most platforms offer an appeal process, and an appeal that adds documentation performs better than one that restates the original complaint.

Suspected violations of the federal rule can also be reported to the FTC at ReportFraud.ftc.gov, which is a separate track from platform removal and is covered in our guide to reporting fake reviews under the FTC rule. Reporting suspected review fraud to a platform or the FTC does not guarantee a review comes down, and review times vary widely by platform and case.

While a report is pending, respond to the review publicly and factually. Readers see the response whether or not the review is eventually removed, and our guidance on handling a negative review online covers what that response should and should not say.

When to Bring in a Reputation Team

Most single fake negative reviews are manageable in-house with the steps above. Businesses can learn how to spot fake reviews and take action. Outside help is worth the investment when the situation has any of these characteristics:

1. Sustained volume across more than one platform, rather than an isolated review
2. A payment demand attached, which shifts the matter into extortion territory
3. Reviews tied to litigation, press coverage, or a public incident
4. Repeated declined reports on reviews where you have documented a clear policy violation

At that point, the work becomes more intensive. It involves continuous monitoring, evidence packages built to platform standards, and escalation through channels that individual owners do not have. These situations also require rebuilding the review profile with legitimate recent feedback so a handful of fakes carry less weight. NetReputation’s review management services and broader online reputation repair work cover all of these angles.

Frequently Asked Questions

Can you tell if a review is fake just by reading it?

Usually not, and less reliably every year. Text-based signals such as grammar errors and exaggerated praise only catch unsophisticated fakes, and research on AI-generated reviews shows they tend to read more cleanly than genuine ones. The dependable signals for how to spot fake reviews are behavioral: posting timing, reviewer history, factual errors about your business, and absence from your customer records.

Are fake reviews illegal in the United States?

Creating, buying, or selling fake consumer reviews is prohibited under the FTC’s Consumer Review Rule, which took effect October 21, 2024 and carries civil penalties of up to $53,088 per violation as of 2026. Our overview of the FTC rule on fake online reviews explains what the rule covers and how to report a violation.

Can a fake review be removed?

Sometimes. Platforms remove reviews that violate their published policies, not reviews a business simply disagrees with. A report that identifies the specific policy breached and supports it with documentation has a significantly better chance than one that does not, though no platform guarantees an outcome, and timelines vary.

Do fake review checker tools work?

Fake review checker tools analyze writing patterns, which makes them a partial tool to detect fake reviews at best. They cannot access your transaction records, your service area, or a reviewer’s posting history, and AI-written reviews are specifically difficult for text-based detection. Treat a checker result as one input alongside your own records rather than as proof.

How common are fake reviews?

Fake reviews are common enough that platforms remove them at industrial scale. Google alone blocked or removed more than 292 million policy-violating reviews in 2025. A separate 2025 analysis of front-page Amazon reviews found roughly 3% were AI-generated, with 93% of those carrying a verified purchase badge.