Manipulated Rank and Fake Signals
Rank can be bought, temporarily. Recognising the signature protects your analysis, and the reasons not to do it yourself are practical rather than moral.
Where a number determines visibility, someone sells a way to move it. Rank manipulation exists, it is detectable, and it distorts analysis that does not account for it.
The methods that exist
Coordinated purchasing. Buying units to drive rank, sometimes with refunds arranged afterwards.
Discount and rebate schemes where buyers are reimbursed outside the marketplace in exchange for a purchase and often a review.
Review manipulation, which is adjacent and covered separately.
Category misclassification to compete in a shallower field.
Keyword manipulation in listings.
Click and traffic schemes aimed at search position rather than sales rank.
The signatures
A rank inconsistent with everything else. Strong rank, very low review count, no advertising presence, unremarkable listing. Products do not usually achieve high velocity invisibly.
A vertical spike with an unnatural shape. Genuine demand events have a build and a decay; a purchased burst frequently appears and disappears more abruptly.
Rank achieved in a category the product does not belong in.
Review bursts clustered in short windows, with similar text lengths and reviewer profiles with narrow histories.
Repeated cycles — spike, decay, spike — at regular intervals, which suggests a purchased schedule rather than organic demand.
A rank-to-review ratio far outside the category norm, in either direction.
Why it matters for analysis
A manipulated competitor distorts your basket and your category median.
Market sizing inflates if it includes purchased volume.
Competitive assessments are wrong. A competitor apparently taking share may be buying position, which is not a durable threat and may be a company about to face enforcement.
Benchmarks become unreachable, and teams are held to targets set against a manipulated comparator.
Exclude suspected products from baskets and note the exclusion. Do not silently drop them.
The practical case against doing it
Beyond it being against every marketplace's terms:
Enforcement is real and improving. Listing suppression, review removal, account suspension. For a seller, the account is the business.
It is not durable. Purchased rank decays exactly like real rank. Sustaining it requires continuous spending with no accumulating benefit.
It corrupts your own data. A seller who buys rank cannot measure anything afterwards, because their baseline is manufactured. Every subsequent test is uninterpretable.
It attracts the wrong customers. Units bought to move rank are not customers; they do not repurchase, refer or review honestly, and the conversion history they create does not represent real demand.
The refund pattern is visible to the marketplace, which reconciles orders and returns in ways an outside observer cannot.
What to do when you suspect a competitor
Nothing, analytically, beyond excluding them from your basket and noting why.
Do not build a strategy around matching them. You would be matching a number that is not sales.
Do not report it as a finding without care. An accusation of manipulation based on rank inconsistency is an inference from weak evidence, and stating it as fact about a named company is a different kind of claim from stating it about the data.
The defensible phrasing: "This product's rank is inconsistent with its review count and advertising presence; we have excluded it from the comparison basket."
The broader point
Any published metric that determines commercial outcomes will be gamed, and the gaming is a cost the honest participants pay in distorted analysis.
The response is not cynicism about the data. It is treating outliers as questions rather than findings, keeping the basket clean, and recognising that a number that is easy to buy is a number to interpret carefully.
Documenting an exclusion
When a product is removed from a basket or a comparison, the removal must be visible.
Record the product, the date and the reason.
State the evidence, in terms of the data rather than as an accusation: rank inconsistent with review count and advertising presence; review burst pattern; rank achieved in an implausible category.
Note the effect on the analysis — how the basket median changes with and without.
Present both where the difference is material, so that a reader can judge.
Do not name manipulation as a fact about a company. The defensible claim is about the data: these observations are inconsistent, so the product has been excluded from the comparison set. That is accurate, it is sufficient for the analysis, and it does not assert something you cannot demonstrate.