ReviewProof ⚖

How should an online store prove its product 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

Every review should be able to name its order. An online store proves reviews by linking them to order records (the one identifier it always holds), sending the request from the delivery event to every fulfilled order, and purging the two habits that poison ecommerce files: the package insert offering a discount for a review, and imported marketplace reviews pasted in without their order data, which land as a wall of UNLINKED rows. At store volumes the monthly import is the whole game — caps are generous (5,000 per import on Pro, 50,000 on Agency) and the signal counts read like operations metrics.

What is online store review provenance?

Online store review provenance is provenance at transactional scale: order-system exports provide reviewer, product, date and the order reference in one file, making ecommerce the easiest sector to link and the easiest to corrupt — because inserts, email flows and third-party review widgets can each inject incentives or gating at a scale no shop counter can match. The audit's job is to reconcile three files that should agree: orders, requests sent, and reviews received.

The insert is an incentive with a print run

The card in the parcel — 'enjoying it? Leave a review for 15% off your next order' — is an incentivised-review programme distributed at fulfilment scale, and it leaves its fingerprints in the review text itself, where customers dutifully mention the offer. ReviewProof's incentive patterns catch the wording; the count, month on month, is the print run measured. Removing the insert and replacing it with the delivery-email ask (no offer, one link, every fulfilled order) is a one-line operations change whose effect the audit then verifies. Stores are often surprised the review rate survives; what falls is the proportion of reviews that read like they were written to claim a voucher, which is rather the point.

The audit mechanics do not change by trade: import the reviews, read the signals traced to your own rows, and publish only what carries a transaction link. What changes is where the risk hides. Service businesses collect reviews at emotional moments — after a repair, a meal, a treatment — and staff under pressure reach for shortcuts: a discount offered at the till, a tablet handed over with the review page already open, a quiet word that unhappy customers should call first. Those shortcuts are precisely the gating and incentive patterns the analysis weights most heavily, and precisely what the FTC rule prohibits. Finding them in your own data first is the cheapest fix there is.

Marketplace reviews: import the data, not the screenshot

Stores selling through marketplaces accumulate reviews whose order data lives in the marketplace's export, in a different column layout, occasionally with dates in a format that will silently defeat burst analysis if unexamined. Import per channel, as files, with the channel's own sample check (one known order, traced from export to report line). Reviews that genuinely cannot be linked — a channel that will not release order references — enter as UNLINKED and stay honest; do not let a widget-scraping tool 'enrich' them with invented references. A store's provenance position is a channel-by-channel table of linked shares, and improving it is a data-engineering roadmap, not a copywriting task.

In practice the routine takes minutes. Export or copy the reviews for the location, import them into ReviewProof by CSV or paste, and read the report top-down: total rows, flagged count, risk score, then the per-signal breakdown. Open the flagged rows themselves — the report traces each signal to its row precisely so you can do this. Decide moderation openly (kept, published, or flagged internally; there is no delete), check which reviews carry a transaction link, and publish or refresh the verified widget so the public claim matches the audited record. On Pro, export the evidence pack while the file is fresh.

Mistakes that undo the work

Importing reviews collected by a widget that itself offered incentives imports the violation wholesale — audit the collection tool before the collection. Discount-code review requests in the delivery email recreate the insert digitally. And de-duplicating 'similar' reviews during spreadsheet preparation, before import, silently edits the duplicate signal out of the evidence the audit exists to examine.

Volume matters differently per business, too. A location with forty reviews cannot absorb a burst the way a location with four thousand can: five same-day reviews are a rounding error in one file and an eighth of the entire profile in the other. ReviewProof detects a burst when at least five reviews land on one day and that day holds at least 40% of the import, which is why the report must always be read against your own baseline. The worked example on this page uses a small file on purpose: small profiles are where provenance work pays back fastest, because every single review is visible.

Worked example: a store finds its insert in its own reviews

A fictional worked example, written to show the numbers and the decisions in practice.

A homewares store imported 2,314 reviews on Pro. The incentive count was 212 — 9% of the file — and reading twenty flagged rows produced the same sentence in different words: the parcel insert's 15% offer, quoted back. The store pulled the insert that week (the replacement card says only thank you, with the support address), kept every row, flagged the 212 internally as a set with the insert named in the log, and moved its whole ask into the delivery email. Incentive counts over the next quarter: 31, 6, 0. Its review rate dipped for one month and recovered; its verified widget, order-linked throughout, now anchors the product pages the ads point to — a use the store describes as the first time its reviews and its orders were the same dataset.

Your five-action checklist

  1. Remove every offer from inserts, emails and widgets connected to reviews
  2. Send the ask from the delivery event to every fulfilled order
  3. Import per channel as files, sample-checking one known order per channel
  4. Track linked share per channel as an operations metric
  5. Reconcile monthly: orders, requests sent, reviews imported

Frequently asked questions

Our review widget collects reviews. Can we import them?

Yes — export from the widget with reviewer, date, text and order reference if it holds one. If the widget offered any incentive for reviews, those rows carry that history; flag the set and fix the collection.

What about reviews on marketplaces we sell through?

Import each channel's export as its own file and check the linked share per channel. What a channel will not evidence stays UNLINKED, honestly.

How many reviews can we import monthly?

5,000 per import on Pro and 50,000 on Agency cover most stores' monthly volume; historical imports split chronologically.

Does the verified widget work for product pages?

The widget displays transaction-linked, kept or published reviews — for a store, order-linked reviews — with certificates behind each, which suits product and collection pages directly.

Keep reading

Related guides from the ReviewProof library:

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.