DigitalMANIPULATIVE

Fake Reviews

What it is

Manufacturing, buying, gating, or suppressing consumer reviews so that a rating summary reflects what the seller wants buyers to believe rather than what buyers experienced.

How it works

Star ratings work because they are read as a census of strangers with no stake in the sale — informational social influence at scale. Fake reviews counterfeit that census. The cheapest forms are direct: reviews written by the seller, bought from review farms, or traded in “free product for five stars” groups. Subtler forms leave every individual review genuine and rig the sample instead: “review gating” invites only customers who first say they were happy, negative reviews are held back or buried, and “brushing” ships unordered goods so a fabricated review earns a verified-purchase badge. Two natural experiments show the incentive at work. Mayzlin, Dover and Chevalier compared hotel reviews on a site that verified stays with one that did not, and found independent hotels sitting next to chain competitors gathered more one-star reviews on the unverified site. Luca and Zervas found roughly one in six Yelp restaurant reviews was flagged as suspicious, with fraud rising when a business's reputation was weak or a competitor arrived. The reader's sense of a consensus is the product; the individual review is only the raw material.

Real-world examples

  • Cure Encapsulations (2019): the FTC's first case against paid fake reviews, brought against a supplement seller that paid a third-party site to write and post Amazon reviews with a target average rating.
  • Fashion Nova (2022): a $4.2 million FTC settlement over review gating — the retailer used a tool that published four- and five-star reviews automatically while holding lower ratings for an approval that never came.
  • The historian Orlando Figes admitted in 2010 to posting pseudonymous Amazon reviews praising his own books and disparaging rivals' — the same mechanism in a market whose currency is reputation rather than sales.
  • Brushing: in 2020 US state agriculture departments traced unsolicited seed packets mailed from abroad to sellers generating verified-purchase reviews for orders that no one had placed.
  • UK consumer group Which? repeatedly documented Facebook groups trading free products for five-star Amazon reviews between 2019 and 2021; the Competition and Markets Authority secured commitments from Facebook to remove such groups and from Amazon and Google to improve detection.

Historical case studies

FTC v. Cure Encapsulations

2019FTC Enforcement

The owner of a supplement company paid a website called amazonverifiedreviews.com to post fabricated five-star Amazon reviews of his garcinia cambogia weight-loss capsules, telling the operator the product needed an average of 4.3 stars to sell and asking him to make it "stay a five star." This was the Federal Trade Commission's first case against paid fake reviews on an independent retail site. The settlement imposed a $12.8 million judgment, suspended on payment of $50,000, and required the company to tell Amazon which reviews it had bought.

Source →

Fashion Nova and the suppressed reviews

2022FTC Enforcement

From 2015 to 2019 the online retailer Fashion Nova used a review-management tool configured to post four- and five-star reviews automatically while holding lower-rated reviews for approval, which the Federal Trade Commission alleged the company never gave. Hundreds of thousands of negative reviews went unpublished. No review on the site was fake; the overall picture was. Fashion Nova paid $4.2 million, in the agency's first case about hiding negative reviews, and ten other companies offering review-management services received warning letters.

Source →

Ethical guidelines

  • Publish every review that meets neutral content rules, in the order and with the weight your interface implies; never gate by sentiment.
  • Disclose any incentive — free product, discount, employment, family — inside the review itself, and never make an incentive conditional on the rating given.
  • The FTC's 2024 rule on consumer reviews attaches civil penalties to fake, purchased, undisclosed-insider, and suppressed reviews; the UCPD (Annex I items 23b and 23c, since 2022) and the UK Digital Markets, Competition and Consumers Act 2024 ban the same practices. Compliance is the floor.
  • Treat a rating system as a measuring instrument: publish the verification method, and let the base rate of one-star reviews show.

How to defend against it

  • Read the distribution, not the average: a bimodal shape (many fives and ones, little in between) or a cluster of five-star reviews from the same week are the classic tells.
  • Sort by most recent and read the one- and two-star reviews first; fabricated praise rarely reproduces the specific complaints real buyers make.
  • Run the listing through an analyser such as Fakespot or ReviewMeta, and search “[product] review” off-platform to read opinions the seller cannot curate.
  • Check the reviewer: profiles that rate dozens of unrelated products in a day, all five stars, in stock phrasing, are review farms.
  • Look for the incentive disclosure; “I received this product for free” is lawful only when stated, and its absence across a run of glowing reviews is itself informative.
  • Report to reportfraud.ftc.gov (US), the CMA (UK), or your national consumer authority under the UCPD (EU); platforms are now liable for taking “reasonable and proportionate steps” to verify that reviews come from real customers.

From the Defense Playbook

Follow the Incentivesminutes

Before weighing a recommendation, find out what the person making it gains if you say yes (commission, funding, votes, attention, status), and adjust how much independent confirmation you require accordingly.

Lateral Readingminutes

To judge an unfamiliar website, leave it: open new tabs and find out what independent sources say about the organization behind it, before spending any time on the site's own content, design, or "About" page.

Reverse Image Searchminutes

Upload or paste an image (or a video frame) into an image search engine to find where else and when it has appeared, which exposes recycled photos, stolen profile pictures, and images given a false caption.

Account Forensicsminutes

Before taking an online account's word, or taking a crowd of accounts as public opinion, spend a minute on the profile itself: its age, history, posting rhythm, network, and photo, which together show whether you are looking at a person, a persona, or a coordinated operation.

Sunlight Against Astroturftakes practice

When a "grassroots" campaign, citizens' coalition, or flood of public comments appears, establish who organized and paid for it and make that public, because manufactured support loses most of its force once the audience knows it is manufactured.

Report Dark Patternsminutes

When an interface tricks you (a hidden charge, a fake countdown, a cancellation maze, consent you never gave), document it and report it to the regulator and the platform, because enforcement against deceptive design is driven by complaint data and one documented report protects people who would never have spotted the trick.

TARES Self-Auditminutes

Before a campaign, advertisement, fundraising appeal, or pitch goes out, test it against the five TARES duties: Truthfulness of the message, Authenticity of the persuader, Respect for the audience, Equity of the appeal, and Social responsibility for the common good.

The Front-Page Testseconds

Ask whether you would be comfortable seeing the tactic, including how it works and why you chose it, described accurately on the front page of a newspaper read by your audience; if the tactic only works when the audience does not know about it, treat that as a finding.

Harm-Class Inventoryminutes

List every persuasive element in a campaign, tag each one as neutral craft, dual-use, or manipulative by design, and for each dual-use element write down where the line is and which side of it you are on.

Disclosure Checklistminutes

Before publishing sponsored, affiliated, incentivized, or endorsed content, check that every material connection between the speaker and the brand or cause is disclosed clearly, conspicuously, in plain language, and in the same place and format as the claim it qualifies.

Every playbook entry states how strong its evidence is and when not to use it. Browse the full playbook.

References

  1. Mayzlin, D., Dover, Y., & Chevalier, J. (2014). Promotional Reviews: An Empirical Investigation of Online Review Manipulation. American Economic Review, 104(8), 2421-2455
    The verified-versus-unverified hotel review comparison showing that competitive incentives produce promotional and negative fake reviews.
  2. Luca, M., & Zervas, G. (2016). Fake It Till You Make It: Reputation, Competition, and Yelp Review Fraud. Management Science, 62(12), 3412-3427
    The estimate that roughly 16 percent of Yelp restaurant reviews are filtered as suspicious and that fraud rises with weak reputation and new competition.
  3. Federal Trade Commission (2024). Trade Regulation Rule on the Use of Consumer Reviews and Testimonials. 16 C.F.R. Part 465, final rule, effective October 21, 2024
    The prohibitions on fake reviews, purchased reviews, undisclosed insider reviews, review suppression, and the civil-penalty authority.
  4. European Parliament and Council (2019). Directive (EU) 2019/2161 amending Directive 2005/29/EC (Unfair Commercial Practices Directive), Annex I items 23b and 23c. Official Journal of the European Union
    The listing of unverified and false consumer reviews as practices unfair in all circumstances, applicable from May 2022.
Last reviewed
Suggest a correction

Detect Fake Reviews in any text

Paste any message, email, or article into our free Manipulation Detector to see if Fake Reviews or other techniques are being used on you.

Related Articles