How do you audit and clean up years of old reviews?
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
Import history as found, split by era so patterns can be dated, judge each era's solicitation by what the record shows rather than by what anyone now wishes, and let every correction carry today's date. Practically: gather the oldest exports first; import chronologically in chunks under your plan cap (500 Free, 5,000 Pro, 50,000 Agency); expect artefacts (drifted dates, lost sources) and note them rather than repairing them silently; flag the clusters each era reveals with the era named in the log; and keep the pre-system history visibly pre-system — its UNLINKED share and missing chains are facts about record-keeping then, not crimes now, provided nothing is backfilled or invented.
What is a historical review audit?
A historical review audit is the retrospective application of provenance discipline to reviews accumulated before any system existed. Its unit is the era — a period with a characteristic solicitation practice, platform mix or management — because the questions history asks are era questions: when did the incentive promotion run, which agency held the account in the burst year, what share of the old file can honestly be linked from surviving order records? The output is a dated map of what the business can and cannot evidence about its own past.
Artefacts are evidence too — do not launder them
Old files arrive damaged in characteristic ways: dates stamped with the export day, sources flattened into 'import', reviewer handles lost. The instinct to repair — re-date the burst artefact from memory, assign plausible sources — must be resisted in its silent form, because a repaired artefact is indistinguishable from a fabricated record. The disciplined version: correct only what a surviving primary record proves (the source export, the order system), note each correction and its basis in the audit record, and where nothing survives, leave the damage visible and noted. An examiner trusts a file more for its admitted gaps; the gaps locate the boundary of what the business claims to know, which is the definition of an honest archive.
Keep the negative reviews. That instruction surprises people mid-crisis, but suppression is itself a violation under the FTC rule, and a profile scrubbed clean during an attack reads as manipulation to anyone who later examines it. ReviewProof has no silent-delete path: moderation decisions are kept, published or flagged internally, each one is logged in the chain of custody, and the evidence export counts the negatives you kept. The strongest response to a suspicious review is a boring, complete record — the row, the dates, the signals, the decision, all preserved while you dispute through the platform's own process.
Judge eras, then close them
Work era by era with a closing note for each: the period, the volume, the linked share achievable from surviving records, the clusters found and their dispositions, the solicitation practice the evidence implies, and the date that practice ended (or the audit's date, for practices ended by the audit). Closing notes convert an embarrassing past into a governed one — the business can answer 'what happened in 2023?' with a document instead of a wince. The final era is always the current system: universal ask, linked imports, monthly packs. Its first closing note is written at its first quarterly review, and from then on history's job is done — the archive's remaining duty is simply to be kept, unedited, beside the living record.
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
Backfilling transaction references for old reviews from plausible guesses is fabrication wearing archive clothing — link only what a surviving record proves. Deleting the worst era wholesale (a spreadsheet tab quietly dropped) is suppression applied to history. And judging past staff by the current policy document, in the log, in personal terms: the audit records practices and dates, not blame.
Triage starts from your own data, not from the public profile. Import the reviews, look at which rows carry which signals, and separate three questions that are often mashed together: is this review about a real transaction, was it solicited in a way the FTC rule prohibits, and what record do we hold if anyone asks? ReviewProof answers the first two from the imported rows — transaction link present or UNLINKED, gating or incentive language present or absent — and answers the third with the provenance chain. What it will not do is declare a review fake: signals are heuristics for human review, and acting as if they were verdicts creates a new problem while solving none.
Worked example: a hotel inherits a decade, maps it in three eras
A fictional worked example, written to show the numbers and the decisions in practice.
A hotel changing ownership imported 1,612 reviews spanning a decade, split into three eras by management: the founding era (412 reviews, 22% linkable from a surviving booking ledger, one burst traced to a launch offer, noted and closed); the agency era (688 reviews, 9% linked, two duplicate clusters and an AI-signal concentration the closing note attributes, on the file's evidence, to the agency's 'content service', contract long ended); and the recent era (512 reviews, 71% linked under the outgoing manager's partial system). The incoming owner did not attempt to purify the file. She priced the eras into the purchase conversation — the agency era's profile was treated as marketing of unknown value — and started the new system on day one with the map as its first document. The map is now the hotel's standard answer when anyone, guest or lender, asks about its oldest reviews.
Your five-action checklist
- Gather the oldest surviving exports and order records before importing anything
- Import chronologically in capped chunks, era by era
- Note artefacts; correct only what a primary record proves, with the basis logged
- Close each era with a written note: volume, linked share, clusters, practices, end date
- Start the current system clean and never edit the archive again
Frequently asked questions
How far back is it worth auditing?
As far as surviving records let you evidence. Era mapping can extend past provable linking — the note for an unprovable era is simply honest about what does not survive.
Can old reviews be linked to transactions now?
Yes, where a surviving record (ledger, order export) proves the match. Linking on that basis is a provenance event, logged with its date — guessing a match is fabrication.
What if the old file contains reviews we believe are fake?
Flag clusters with their pattern and era logged, dispute through the platform where worthwhile, keep the rows. The historical record stays complete.
Does an ugly history hurt the current business?
An unmapped one does. An era-mapped history with the current system running clean is evidence of governance — the quality buyers and lenders actually price.
Keep reading
Related guides from the ReviewProof library:
- What should you do if a competitor is leaving fake reviews?
- What should you do in the first 48 hours of a one-star attack?
- How do you clean up a profile with bought reviews in its past?
- Why are employee reviews of your own business a violation risk?
- What does it mean when one reviewer appears at many businesses?
- What should you do when a customer threatens a bad review?
- Can having too few negative reviews look suspicious?
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.