Together

Consumer investigation

Fake Reviews, False Origins:
Don’t Reward Marketplace Cheats.

Amazon banned hundreds of Chinese brands over repeated policy violations, especially review abuse. Walmart investigations exposed counterfeit goods, stolen seller identities, and questionable origin claims. A star rating is not a trust certificate.

Why you can't trust ratings on their own

A posted star average is review feedback, not an independent test of product quality. Bought or incentivized reviews can inflate it, and reviews can refer to a different seller or product variant. Even genuine reviews do not establish long-term durability or verify a manufacturing claim.

Read the evidence below before treating stars as a purchase verdict. A rating without independent corroboration is unverified, not automatically fake. Together keeps its environmental grade, the Product Confidence reputation pilot and the posted marketplace rating separate.

Buying positive reviews is cheating. Paying people to attack a competitor is cheating. Using false manufacturing claims to win a shopper’s money is deception. None becomes acceptable because the product is cheap—or because a familiar marketplace processes the payment.

For shoppers trying to reward honest businesses, those practices are not a side issue. Conscious shopping means asking who deserves the sale—not just which listing has the most stars.

Together’s position is straightforward: sellers should earn your money with products and claims they can substantiate, not purchased credibility. The evidence gives consumers concrete reasons to scrutinize sellers, ratings, and origin claims before buying.

The Chinese-brand crackdown was substantial

In September 2021, Amazon confirmed to The Verge that it had permanently banned over 600 Chinese brands across 3,000 seller accounts following a five-month global crackdown. Amazon attributed the bans to knowing, repeated, significant violations of its policies, especially those concerning review abuse.[1]

That is not an anecdote about one questionable listing. It is a large enforcement action involving hundreds of brands. The report discussed familiar electronics names, including Aukey, Mpow, and RavPower, and tactics ranging from gift cards for reviews to rewards for deleting negative feedback.

Graphic 01 · documented enforcement

Hundreds of Chinese brands. Thousands of accounts.

600+Chinese brands banned
3,000Seller accounts covered
Amazon’s reported total after a five-month global crackdown, disclosed September 2021. Brands and accounts are different units; neither figure counts fraudulent products or reviews.

Source: The Verge, reporting Amazon’s confirmation. These are historical bans; this article does not establish every brand’s current selling status.

Chinese operators also appear in Amazon’s later enforcement account. Its February 2024 update reports action during 2023 against 61 bad actors in China, including 13 new lawsuits and two criminal referrals. The same account reports action against 58 bad actors in the United States and 44 in Europe and the United Kingdom.[2]

China has the largest count in that particular summary. But the figures do not answer which country has the highest fraud rate. They count targets of enforcement, and the source does not provide comparable totals of sellers or reviews examined in each geography. Calling those counts a fraud-rate ranking would misstate the data.

Graphic 02 · same metric, different reporting groups
These counts describe Amazon’s reported enforcement activity, not random sampling, national prevalence, or convictions. The groups include two countries and one combined region.
Exact chart data
Geographic reporting groupActors targeted in 2023
China61
United States58
Europe / United Kingdom44

Source: Amazon, February 7, 2024 update. No aggregation beyond the source’s reporting groups.

What research says about bought reviews and product quality

Enforcement tallies show who got caught. A peer-reviewed study asks a different question: what kind of products buy reviews, and what happens to ratings when the buying stops?

Graphic 02A · peer-reviewed research, observed review-buying sample

Where bought reviews were linked, and what happened afterward

84%China-based sellers

Among product-level observations with identified seller location in the observed fake-review sample. Not a count of distinct companies or individual reviews.

The study tracked roughly 1,500 Amazon products soliciting reviews in selected Facebook groups, Oct 2019–Jun 2020. It is not a sample of all Amazon products.

≈70%relative increase in one-star review share

Measured after review recruiting stopped. Relative change, not 70 percentage points.

It does not mean 70% of products failed. Later negative text emphasized quality, breakage and value.

  • Not84% of all fake reviews bought by Chinese companies.
  • Not84% of Chinese companies buying reviews.
  • Notmanufacturing origin: seller location is not where goods were made.
  • Notlaboratory quality tests: quality is inferred from consumer feedback.
Sample is a selected, observed review-buying market, not all Amazon products. Numeric details from the December 2021 author-hosted version (Table 3, Section 6).

Source: He, Hollenbeck and Proserpio, The Market for Fake Reviews.

The study’s published conclusion is consistent with review manipulation being used mainly by low-quality products. Documented review buying is a quality warning, not just an ethics problem.

The authors infer quality from what buyers later wrote, and they acknowledge limits in using ratings this way. That supports a real concern that fake reviews conceal poor products. It does not prove the motive or quality of any single seller, and a China-based location is no verdict on any company.

So the practical advice stands: look for independent tests of the exact product, check warranty and support terms, and check the sources behind claims. Stars alone are not enough to establish quality, seller honesty or manufacturing origin.[7]

These operations sell an appearance of genuine demand

A review does not have to come from a bot to be manipulated. A real person can buy a real product and then be reimbursed for posting the review the seller wants.

Amazon’s account of the Tahoevine network describes sellers paying fees to websites that offered reviewers full refunds in exchange for five-star reviews. Its August 2023 description of a lawsuit against PMNLWeb goes further: reviewers allegedly received instructions on searching for and purchasing products so the transaction would look like an ordinary shopping journey. Reimbursement followed the positive review.[2]

Graphic 03 · process explainer, not a measured flow

How a paid review can resemble a normal purchase

  1. 01Seller pays a brokerThe seller buys access to a review service or refund network.
  2. 02Reviewer buys the productA genuine transaction creates the appearance of an ordinary customer purchase.
  3. 03Reviewer posts positive feedbackThe favorable review is the condition for the reward.
  4. 04The purchase is reimbursedThe reviewer recovers the purchase price outside the normal sale.
Simplified schematic of conduct described in Amazon’s Tahoevine and PMNLWeb accounts. The PMNLWeb description concerns allegations in an August 2023 lawsuit, not a court finding reviewed for this article. No transaction volumes are represented.

Source: Amazon’s enforcement account.

The practical consequence is uncomfortable but simple: a purchase can be real while the endorsement is bought. Purchase verification cannot, by itself, establish that the reviewer was independent.

The manipulation can also run in the opposite direction. Amazon reports that a China-based broker identified as Mr. Wu sold both positive reviews to raise clients’ product rankings and negative reviews to lower competitors’ rankings. Amazon says local authorities confirmed the activity and that he received a criminal sentence in China. That is a reported criminal outcome, distinct from Amazon merely filing a complaint; this article relies on Amazon’s account, not an independently reviewed judgment.[2]

The conduct is not limited to China. In April 2023, the FTC approved a final consent order requiring The Bountiful Company to pay $600,000 in consumer relief over allegations of review hijacking: repurposing other products’ reviews and ratings to promote supplements. A settlement is not a criminal conviction, but it is a documented regulatory outcome.[3]

The FTC’s 2024 final rule also addresses compensation conditioned, explicitly or implicitly, on positive or negative review sentiment. The relevant distinction is the conduct and the condition attached to the incentive—not a blanket claim that every incentivized review is illegal.[4]

Walmart’s logo is not a guarantee of seller identity or authenticity

In a September 2025 investigation, CNBC identified at least 43 Walmart Marketplace vendors that used another business’s identity to set up their accounts. It also tested 20 products offered by impersonating sellers. All 20 were determined to be counterfeit.[5]

That sample consisted of six items CNBC purchased and 14 items customers had bought and returned to the business being impersonated. Brands authenticated products in most cases; some were assessed through laboratory comparison with authentic products.

Graphic 04 · targeted investigation

What CNBC found on Walmart Marketplace

43+

Vendors identified using another business’s identity.

20 of 20

Tested products from impersonating sellers determined to be counterfeit.

Not a random marketplace sample. The 20 products were selected from suspect sellers; these results cannot be extrapolated into a counterfeit rate for all Walmart purchases. The investigation does not establish Chinese manufacture of all tested goods.

Source: CNBC, September 19, 2025.

Those results support a specific conclusion: Walmart’s marketplace controls failed to keep identified impersonators and counterfeit products away from customers. They do not support saying every Walmart seller is dishonest or that every low-cost product is counterfeit.

Walmart told CNBC it has a zero-tolerance policy for prohibited or noncompliant products. CNBC also reported that Walmart tightened some health-and-beauty seller vetting after receiving the investigation’s findings. Those responses matter, but they do not erase the documented failures.

A high aggregate rating can hide the experience that matters

CNBC’s February 2025 review of a Walmart listing for Sol de Janeiro’s Brazilian Bum Bum Cream found a headline score of 4.6 out of five stars, based on 2,526 ratings. The listing had 1,552 written reviews, but only 246 were identified as verified purchases from Walmart. Of those 246, 118 were one star—47.97%.[5]

Graphic 05 · one listing, one historical snapshot
Headline rating: 4.6 / 5 · 2,526 ratings · 1,552 written reviews
The pie shows only the 246 verified-purchase written reviews, not the 2,526 ratings or all 1,552 written reviews. The listing pooled reviews across sellers. “Other star ratings” is the remainder: 246 − 118 = 128.
Exact chart data and calculation
Verified-purchase reviewsCountShare
One star11847.97%
Other star ratings12852.03%
Total246100%

Source: CNBC, February 2025 snapshot reported September 2025. Percentages calculated from its counts.

This discrepancy is not proof that unverified reviews were fabricated. Nor does a one-star complaint, on its own, prove counterfeiting. It demonstrates why an overall product score can be a poor substitute for checking which seller supplied the item and what recent buyers of that exact version experienced.

A good review of a genuine product bought from one seller does not authenticate a unit offered by another seller. That is a marketplace trust problem even when the review itself is honest.

Product origin lookup: verify the manufacturing claim

In July 2025, FTC staff wrote to Walmart about potentially misleading “Made in USA” claims by third-party sellers. The letter cited flag listings advertising American manufacture while their specifications described the items as “imported.” It also identified a China-based seller offering whiskey glasses with an American-made claim.[6]

The contradictions deserve scrutiny. The seller’s location alone, however, does not establish where the glasses were made. The letter warns about potentially unlawful claims requiring substantiation; it expressly does not assess whether Walmart itself violated the FTC Act or the Made in USA Labeling Rule.

For an unqualified “Made in USA” claim, the FTC’s standard is “all or virtually all” made in the United States. An American-sounding brand, U.S. shipping address, or domestic warehouse does not independently demonstrate that standard has been met.

Keep three facts separate: where the seller operates, where a review broker operates, and where the product was manufactured. The Chinese-brand crackdown establishes misconduct involving Chinese brands. It does not establish the manufacturing origin of every product those brands sold. Specific product accusations need model-level evidence.

Before you buy, make the seller earn your trust

You do not need to prove a crime before deciding not to buy. An unresolved seller identity, conflicting origin claims, or an offer of money for favorable feedback is a practical reason to stop and verify. The aim is not to guess guilt from a country label. It is to avoid funding conduct and claims that cannot withstand scrutiny.

Graphic 06 · shopper verification checklist

Four separate checks. No single trust shortcut.

01 · seller identity

Who gets your money?

Open the actual seller profile, not just the brand page. Compare the legal business details with the manufacturer’s authorized-seller information. A marketplace logo is not the seller’s identity.

02 · manufacturing origin

Where was this exact model made?

Look for model-specific labeling, packaging, and manufacturer documentation. Distinguish “designed in,” “ships from,” and “assembled in” from an unqualified manufacturing claim.

03 · independent evidence

Who checked the product?

Seek testing or authentication that identifies the model, method, and source. Check relevant certification or recall records where applicable. Affiliate recommendations and star ratings are not laboratory results.

04 · review integrity

Is the feedback about this purchase?

Read recent reviews for the exact version; look for unrelated product references and rewards tied to positive feedback or removing criticism. Patterns prompt investigation—they do not prove fraud.

Together’s editorial checklist, informed by the documented cases in this article. These are verification steps, not a validated fraud-detection model or a guarantee of authenticity.

If a seller offers a refund or gift card conditioned on a positive review or removal of criticism, preserve the message and report it through the marketplace. For suspected fraud in the United States, the FTC accepts reports at ReportFraud.ftc.gov. A refund may resolve your purchase; it does not make the practice acceptable.

For health and beauty products, an unverifiable seller or doubtful authenticity is a reason to choose an authorized source rather than gamble on a discount. For domestic-origin purchasing, require evidence for the origin claim rather than rewarding patriotic branding.

Stars are feedback—not proof of origin, authenticity, or an honest seller.

An ethical shopping assistant should help you ask better questions—not replace one unexplained rating with another. Together’s ethical shopping browser extension brings product-origin information, environmental-impact information, and alternative products into the shopping process. That is separate from certifying review authenticity: this article does not claim that Together detects fake reviews or guarantees a seller’s honesty.

The standard should be demanding: identify who is selling, verify what can be verified, and make the uncertainty visible. Do not reward a seller merely because manipulated signals make the offer look credible.

Sources and evidence notes

Research snapshot prepared September 30, 2026; updated October 6, 2026 with the peer-reviewed review-buying study. Dates below describe the evidence—not a claim that every condition remains unchanged today. This is a sourced editorial analysis, not an independent product-testing investigation.

  1. The Verge: Amazon says it’s permanently banned 600 Chinese brands for review fraud

    September 17, 2021. Includes direct Amazon confirmation of over 600 brands, 3,000 seller accounts, and the stated grounds. Platform enforcement is not a court judgment; current selling status is not verified here.

  2. Amazon: How Amazon takes action to stop fake reviews

    Published June 2023; updated February 7, 2024. Includes 2023 geographic enforcement totals, Tahoevine and PMNLWeb descriptions, and the reported Mr. Wu criminal outcome. Corporate account by a party to the enforcement actions. The article does not independently inspect the underlying court judgments or update the civil dockets.

  3. FTC: Final order against The Bountiful Company

    April 10, 2023. Final consent order requiring $600,000 in consumer monetary relief and restrictions on deceptive review tactics. Underlying conduct was alleged in the complaint; a consent resolution is not a criminal conviction.

  4. FTC: Final rule banning fake reviews and testimonials

    August 14, 2024 announcement. Describes sentiment-conditioned incentives, false reviews, certain undisclosed insider reviews, and other prohibited practices. This article does not offer a legal opinion on any unnamed seller.

  5. CNBC: Walmart’s Marketplace boom—lax vetting, identity theft and fakes

    September 19, 2025 investigation. At least 43 impersonating vendors; 20 counterfeit products in a targeted sample; February 2025 Brazilian Bum Bum Cream review snapshot. CNBC conducted the reporting and authentication work. Its sample is not a marketplace-wide prevalence estimate. Reviews pooled across sellers; unverified does not mean fake.

  6. FTC Bureau of Consumer Protection: Letter to Walmart on misleading Made in USA representations

    July 8, 2025. Staff warning about potential third-party seller violations, with listing examples. Not an adjudication of each product’s manufacturing origin and expressly not an assessment of Walmart’s legal liability.

  7. He, Hollenbeck and Proserpio: The Market for Fake Reviews

    Marketing Science 41(5), 896–921 (2022), DOI 10.1287/mksc.2022.1353. Numeric details (84% China-based seller location, counted at product level; about +70% relative rise in one-star share) are from the author-hosted December 2021 version, Table 3 and Section 6; the 2022 published abstract supports the low-quality conclusion. Observed sample, not all Amazon products; quality inferred from reviews, seller location is not manufacturing origin.