Online Review Statistics for 2026: What Readers Expect, What Platforms Remove, and What the FTC Now Fines

Twenty review figures with the organization that counted each one

Every figure below was opened on the publisher's own page in October 2026. The 2016 and 2017 rows are flagged as older. The last column says who was asked and whether the publisher sells something related. Local search behavior beyond reviews lives on our local SEO statistics page, and the rest of the section is on the marketing statistics hub.

The star floor moved up half a point in a year

BrightLocal has run its Local Consumer Review Survey for years, and its February 2026 edition, from a SurveyMonkey panel of 1,002 US adults, shows standards tightening fast. 92% say they care about star ratings. The share who will only use a business rated 4.5 or higher nearly doubled, from 17% to 31%. One in ten will accept nothing below five stars, which is a hard standard for any business that has ever had a bad Tuesday.

Recency tightened too. 32% now look for reviews from the last two weeks, up from 20%, and 18% are swayed only by reviews from the last week. A profile with a strong average and nothing new since spring now reads as a business that may have stopped trading, or stopped asking. The effect cuts both ways: 85% say positive reviews make them more likely to use a business and 77% say negative ones make them less likely.

For nearly a third of customers, a 4.3 is now a reason to keep scrolling.

Money follows the reading. In the same survey, 93% said they had bought something after reading reviews, and 27% had spent over $1,000 on that basis. 70% said they had regretted at least one purchase made after reading reviews, which suggests the reviews were persuasive more reliably than they were right.

Why a perfect 5.0 is not the number to chase

The best independent evidence on ratings and sales is now old. Northwestern University's Spiegel Research Center, in work first published in 2017, found that purchase likelihood for a product with five reviews was 270% greater than for one with none, and that the benefit of more reviews faded quickly after the first five. Purchase likelihood typically peaked at ratings between 4.0 and 4.7 and declined as ratings approached a perfect 5.0. Displaying reviews mattered more for expensive items: conversion rose 380% for a higher-priced product, against 190% for a lower-priced one.

It is from 2017 and about products, not plumbers, so read it as a direction rather than a forecast. We include it because it is a university study with no review software to sell, and because nothing newer from an independent source measures the same thing. Pew's December 2016 report found 48% of review readers said it is often hard to tell whether reviews are truthful and unbiased.

Readers grade the reply as closely as the review

The owner's reply has become part of the product. BrightLocal's 2026 survey found 89% expect owners to respond, 80% are likely to use a business that responds to all reviews, and 42% are unlikely to use one that never replies. Speed expectations jumped: 19% expect a same-day reply, up from 6%, and 32% expect one by the next day, up from 18%. 81% expect a reply within a week.

The catch is the robot reply. Half the panel said generic or templated responses put them off. That matters now that most review tools offer AI drafts. A draft is fine. A draft posted unread, thanking a customer for their kind words under a one-star complaint, is the kind of reply that ends up in a screenshot. Our guide to reputation management tools covers which reply features to switch on, and AI quotes a bad review covers the reply that has to work hardest.

Who writes reviews, and how often someone asked first

Most reviews are invited. BrightLocal found 78% of consumers were asked for a review in the past year, and 28% say they will always write one when asked, up from 16%. 69% wrote at least one business review in the past 12 months, and 60% have shared a positive experience against 29% a negative one. BrightLocal puts the average at four to six reviews a year per person.

• Where they post: Google 45%, Facebook 34%, Yelp 24%, Apple Maps 17%, Tripadvisor and BBB 16% each.
• 59% were offered something before leaving a review, down from 64% the year before.
• 94% say they are open to writing reviews; only 6% rule it out.
• 11% were offered an incentive specifically for a positive review, which is the practice the FTC rule bans outright.

The lesson for an owner is that asking works and buying is unnecessary. If 78% of your customers are being asked by someone, the business that never asks is not being modest; it is being outvoted. The guide to getting more Google reviews has the link, the scripts and the rules side by side.

Fake reviews, by the platforms' own count

Every major review platform now publishes an annual enforcement number. None is audited, and each measures something slightly different, so read them as the platform's account of its own work rather than as a census of fakes.

The rising totals do not prove fakes are multiplying. They prove detection is, and the platforms say as much: Google credits its models, Yelp says it upgraded its recommendation software to catch thin and conflicted reviews. On Tripadvisor, review boosting by owners, staff or their friends made up 54% of the fraud it caught in 2024.

The FTC rule, in dates and dollars

The FTC approved its final rule on fake reviews and testimonials by a 5 to 0 vote on August 14, 2024, and the agency's questions and answers say it went into effect on October 21, 2024. It bans fake reviews, including AI-generated ones; paying for reviews conditioned on a positive or negative sentiment; undisclosed reviews by insiders; company-controlled sites posing as independent reviewers; suppressing reviews through groundless legal threats or intimidation; and buying fake social media indicators. Courts can impose civil penalties for knowing violations.

1. 1

   October 21, 2024

   The rule takes effect. Ordinary consumers cannot be liable for what they write; the rule's content provisions apply to businesses.

2. 2

   December 22, 2025

   The FTC sends warning letters to 10 unnamed companies, citing civil penalties of up to $53,088 per violation. The agency says the letters are not findings that anyone broke the rule.

3. 3

   May 11, 2026

   The FTC and Illinois announce a civil penalty case, filed by the Department of Justice, against a company they allege built thousands of fake home repair listings with fabricated five-star reviews. The complaint cites the review rule among other laws. These are allegations, not findings.

Consumers would go further than the rule does. In BrightLocal's 2026 survey, 97% said businesses caught with fake reviews should face some punishment, 57% favored a ban from review platforms, 37% fines and 16% criminal charges. Note what the rule does not cover: it has no specific ban on asking only happy customers, though FTC staff say that can still violate the FTC Act, and Google bans it outright. DataCram is not a law firm; for a business in a regulated trade, counsel should approve the review program.

AI summaries now read the reviews for the customer

The newest number in this field may matter most. In BrightLocal's 2026 survey, 82% said they read AI-generated review summaries. Of those, 23% would rely on the summary alone, 39% would read it alongside the written reviews, and 18% of the whole panel skip the summaries. 40% said they trust AI platforms for business recommendations and 32% said they do not.

So a growing share of customers never read your reviews at all. They read a machine's paragraph about your reviews, which is built from the patterns in what customers wrote. Three complaints about late arrivals become a sentence. How each engine uses review text is explained in reviews and AI recommendations, and the free AI visibility check shows which businesses ChatGPT and other engines name for your customers' questions.

How to read these numbers before one ends up in a sales deck

1. 1

   Notice who sells review software

   BrightLocal, the source of most consumer figures here, sells review and listing tools. Its panel and questions are published, which is more than most, and its conclusions were never going to be that reviews are overrated.

2. 2

   Separate saying from doing

   Survey answers are self-reports. People who say they only use 4.5-star businesses still call the 4.2 with the fastest reply. Platform counts measure behavior, but only the platform's own.

3. 3

   Treat enforcement totals as report cards the platforms wrote

   A bigger removal count can mean more fakes, better detection or a broader definition. No outside auditor checks these figures.

4. 4

   Check the page for its own contradictions

   BrightLocal's survey page gives two figures for how many people asked for a review wrote one, 65% and 83%. When a source disagrees with itself, quote neither or quote both with the conflict.

5. 5

   Date every figure

   The Spiegel and Pew numbers are from 2017 and 2016. They are here because nothing independent replaces them, not because review habits stood still.

The four review numbers you can measure yourself this week

National surveys set expectations. Your own profile tells you whether you meet them. Count four things: new Google reviews divided by completed jobs, your average rating over the last 90 days rather than all time, the age of your newest review, and your median reply time. Then compare all four with the three businesses ranked above you in the map pack, which is exactly what our free map pack audit does.

The local visibility check shows where you appear in Google search and maps for your customers' terms, and the local SEO tools guide compares the software that automates requests and replies. When you would rather have it run for you, review generation and replies are part of DataCram's SEO programs, which start at $575 a month, and our local marketing plans start at the same monthly price. On sites built with Athena CMS, the Echo agent drafts replies for a person to approve and flags unhappy reviews first.