Online reviews substantially influence purchasing, which creates a strong incentive to manipulate them. The manipulation has recognisable patterns.
The forms it takes
Several distinct practices.
Paid reviews, where a seller compensates somebody for a positive review, frequently arranged through private groups and messaging apps.
Incentivised reviews, where a free or refunded product is provided in exchange, which platforms generally prohibit for that reason.
Review hijacking, where an established listing with many positive reviews is edited to sell an unrelated product, transferring the reputation.
Review merging, where variations of a product share a review pool, so reviews of one item appear under another.
And negative manipulation, where competitors post poor reviews.
The regulatory position
Which has been strengthening.
Consumer protection authorities in several jurisdictions have brought enforcement actions against sellers and against facilitators of fake reviews.
Rules prohibiting fake reviews and undisclosed incentivised reviews have been introduced or clarified in several places, with penalties.
Platforms have invested in detection and have removed large numbers of accounts and reviews, and have pursued litigation against brokers.
The problem persists because the incentive is strong and the operations move quickly.
The signatures worth recognising
What to look at on a listing.
A large number of reviews clustered in a short period, particularly shortly after listing.
Reviews that describe the product in marketing language rather than in use.
Reviews mentioning a different product entirely, which indicates merging or hijacking.
A rating distribution that is heavily weighted to five stars with almost nothing in between, which is unusual for genuine distributions.
Reviewer profiles with a large number of reviews across unrelated categories in a short period.
And a product with many reviews and a very short listing history.
What the middle reviews tell you
The most useful single technique.
Three-star reviews are the least likely to be manipulated in either direction and are generally the most informative, since they describe both what works and what does not.
Sorting to recent also helps, since it reveals whether quality has changed, which happens when a seller changes supplier.
And reading the negative reviews specifically for the failure mode rather than the tone identifies whether the problem is one you would care about.
The verification analysis tools
Third-party services exist that analyse review patterns and estimate what proportion appear unreliable.
They are heuristic rather than definitive and they identify obvious manipulation reasonably well.
Using one as a filter rather than as a verdict is the reasonable application, and a listing flagged as heavily manipulated is worth avoiding regardless of the precision of the estimate.
Where reviews are more reliable
Sources with structural reasons to be better.
Retailers requiring verified purchase, which raises the cost of manipulation without eliminating it.
Independent testing organisations that buy products and test them, funded by subscription rather than by advertising or affiliate revenue.
Specialist community forums where participants have long histories and reputations within the community.
And professional review sites that publish their testing methodology, which allows the assessment to be evaluated rather than taken on trust.
The affiliate disclosure question
Worth applying to any recommendation.
A substantial proportion of published product recommendations carry affiliate commission, which is legitimate and is required to be disclosed in many jurisdictions.
The relevant question is not whether commission is earned but whether the recommendation would change without it.
Sources that recommend against buying things, that name specific faults, and that occasionally conclude nothing in a category is worth buying are demonstrating independence in a way that a source recommending a product in every article is not.
What I actually do
Read three-star reviews first.
Sort to recent.
Check the rating distribution shape rather than the average.
Look for one independent test of the specific model.
And treat a product with only enthusiastic reviews and no criticism as a warning rather than a recommendation, since genuine products accumulate complaints.
Writing useful reviews
Worth a note since the whole system depends on people bothering.
The reviews that help are specific — what you used it for, what worked, what did not, and how long you have had it.
Updating a review after several months is disproportionately useful, since durability is the thing initial reviews cannot address and is what buyers most want to know.
And describing a fault precisely, rather than expressing frustration, is what makes a negative review actionable for somebody deciding.
Question and answer sections
An underused source on marketplace listings.
Answers come from previous buyers and from the seller, and buyer answers are frequently more candid than reviews because they are responding to a specific question.
They are also less targeted by manipulation, since the format does not carry a rating.
Reading them, particularly the questions about compatibility, dimensions and what is included, resolves a substantial share of the uncertainty that reviews do not address.
Photographs in reviews
Consistently the most useful element and the hardest to fake convincingly.
Buyer photographs show the actual product, at actual scale, in ordinary lighting, which resolves questions about colour, size and finish that no description does.
They also reveal packaging and condition, which for marketplace purchases is informative about the seller.
Filtering to reviews with images, where the platform allows it, is a fast way to reach the informative subset.