What is a reviewer fingerprint?
By the Ingenious AI team · Published 14 November 2026 · Reviewed 14 November 2026
Run the free review audit See pricing · See a sample provenance report
Short answer
A reviewer fingerprint is a salted HMAC-SHA256 hash of the reviewer's normalised handle and source — a stable pseudonym that contains no readable identity. ReviewProof pools only the hash, counts and flag state across businesses: if the same fingerprint was flagged at another business, it adds risk at yours (15 points when flagged elsewhere, 5 when merely present), shown as counts only, never as another merchant's name, identity or reviews. It is how the fraud graph spots a travelling fake reviewer without building a dossier on anyone.
What is a reviewer fingerprint?
A reviewer fingerprint is a one-way, salted hash that lets a system recognise 'the same reviewer handle on the same source' across many businesses without storing or sharing who that reviewer is. One-way means the hash cannot be reversed into the handle; salted means the same handle hashes differently outside this system. Pooling hashes and counts — never raw identities or review text — is the privacy boundary the graph is built on.
Why pooling counts, and only counts, matters
A fake-review seller's one structural weakness is reuse: the same stable of handles reviews business after business. A single business sees one plausible row; the pool sees the handle's ninth flagged appearance. But a pool of raw identities would be a surveillance asset and a liability. ReviewProof's compromise is strict: what travels is the fingerprint, a review count, and a flagged state with a reason category. What never travels is the identity, the review text, or which merchant flagged it. You learn 'this fingerprint is flagged elsewhere' — enough to weigh the row in front of you, not enough to learn anything about anyone else's customers.
ReviewProof computes every signal from the rows you import, never from a reputation guess about your business. The risk score is 100 × (gating × 1.0 + incentive × 1.0 + duplicate × 0.8 + AI-signal × 0.6) ÷ (2.6 × number of reviews), plus 15 points when a burst is detected, capped at 100. Each flagged review in the report names the imported row that triggered the signal, so a human can open that row, read it in context, and decide. The product labels these patterns signals for human review, not verdicts — and that restraint is deliberate, because a heuristic that overclaims would be its own kind of false evidence.
What to do when a fingerprint is flagged elsewhere
Treat it as context, not conviction. The row in front of you might be that reviewer's one honest visit. Check the row's own evidence: transaction link, wording signals, timing against any burst. If the row is UNLINKED and carries wording signals and the fingerprint is flagged elsewhere, the combined picture justifies flagging it internally and noting the combination in the record. If the row is transaction-backed with specific, ordinary wording, the honest decision may be to keep it and note why. Either way the reasoning is logged — the graph informs judgement; it is deliberately not built to replace it.
ReviewProof's import is forgiving about format and strict about meaning. Column names are matched flexibly — body, review or text for the review itself; rating or stars; reviewer, author or name; a verified flag and a transaction reference when you hold them. Rows are capped per import by plan (500 on Free, 5,000 on Pro, 50,000 on Agency), which is generous for a single location's history. What the importer will not do is upgrade a review's status: a row that arrives without a transaction link is analysed fully and still labelled UNLINKED, because the label is the product.
Mistakes that undo the work
Trying to reverse or guess a fingerprint's identity defeats the design and the point — work with the count and the flag. Treating 'present elsewhere' (5 points) as suspicious in itself is another error: active genuine reviewers appear at many businesses, which is why presence weighs a third of a flag. And the graph is a Pro and Agency feature — do not promise its pooled view on a Free workspace, where only your own file's analysis runs.
Two design choices keep the detection honest. First, nothing is scored until you import it: there is no scraping of your public profiles and no third-party score blended in. Second, the weights reflect what the FTC fake-review rule (16 CFR Part 465) actually prohibits — how a review was solicited weighs heaviest (gating and incentive markers at 1.0 each), while wording patterns weigh less (duplicates 0.8, AI-generation patterns 0.6). A burst of same-day reviews adds a flat 15 points because timing is visible in the dates you supply. You can reproduce every number in the report from your own file, which is exactly the point.
Worked example: the handle that reviewed eleven businesses
A fictional worked example, written to show the numbers and the decisions in practice.
A restaurant group on the Agency plan imported 392 reviews across three locations. Four rows shared a fingerprint that the pooled graph reported as flagged at other businesses. Individually, two of the four looked unremarkable. Together — same handle and source, flagged elsewhere, all four UNLINKED, three inside one burst week, two carrying AI-phrase wording — they formed a pattern no single location could have seen. The group flagged the four internally, logged the combined reasoning (fingerprint status plus links plus timing, each element named), and included the cluster in a platform report. Another merchant's private data was never visible at any point — only counts and flag state.
Your five-action checklist
- Import with consistent reviewer handles and source values so fingerprints match
- On Pro or Agency, check flagged-elsewhere counts in the graph view after each import
- Judge flagged-fingerprint rows on the full combination: links, wording, timing
- Log the combination you relied on when you flag a row internally
- Never attempt to identify a fingerprint or ask what another merchant flagged
Frequently asked questions
What exactly is hashed?
The reviewer's normalised handle together with the source, using HMAC-SHA256 with a salt. Raw identities are never pooled — only the hash, counts and flag state.
How much does the graph add to risk?
A fingerprint flagged at another business contributes 15 points; a fingerprint merely present elsewhere contributes 5. Both are shown as counts only.
Can I see which business flagged a reviewer?
No. The pooled layer exposes counts and flag state only — never another merchant's identity, reviews or store data. That boundary is the design.
Which plans include the reviewer-fraud graph?
Pro and Agency. Free covers one location with the full risk analysis of your own imported file, certificates and the verified widget.
Keep reading
Related guides from the ReviewProof library:
- How can you prove a review is real?
- How do you detect fake reviews in your own profile?
- What is review gating, and how do you stop doing it?
- What counts as an incentivised review?
- What does duplicate review text tell you?
- What is a review burst, and when is it innocent?
- Can you tell if a review was written by AI?
Start with the tool itself: open the free audit desk, see how ReviewProof works, compare Free, Pro and Agency plans, inspect the sample provenance report, or browse all ReviewProof guides.