Reviews

“Verified” is a spectrum, so we show where on it each review sits.

A single verified badge hides how much proof is actually behind a claim. Someone who confirmed a work email and someone who connected their live account are both “verified” and are not remotely the same evidence. Every review here carries its tier, its provenance and its computed weight, on the card.

Five tiers

What each one proves, and what it is worth

Tier	How it is earned	How it appears and counts	Weight
Registered	Email sign-up only	Published in full, marked, and counted at zero. May comment and vote.	0.0
Identity verified	LinkedIn or work email on a real company domain	Standard review. Counted in the score at full tier weight.	1.0
Verified user	In-product screenshot, invoice, or vendor-confirmed customer match	Prominent 'Verified user' label. Counted at 1.6× a standard review.	1.6
Verified active customer	Signed in with the product itself, confirming an active account	'Verified active customer'. Counted at 2.2×, the highest user weight.	2.2
Expert tester	Hands-on test by named Launch500 staff	Shown separately as the editorial verdict. Never blended into the community average.	0.0

Integration verification confirms an active account without sharing any data from it. We receive a yes or a no and nothing else. We never see the reviewer’s usage, and we never disclose a reviewer’s identity to a vendor, including when the vendor disputes their review.

Moderation

What happens between submit and publish

Automatic checks, a hold, then publication with the whole Crew watching. A person steps in only where something is reported, which is how a small team keeps every review honest.

01
Automatic checks at submit
Pros and cons must each say something substantive. Links and contact details are refused, so a review cannot become an advert or expose anyone. A review written while signed in at the product's own domain is refused, and each account can submit ten a day.
02
A hold before anything appears
Every review waits at least 48 hours, and a new account's reviews wait until the account is three days old. A burst of fresh accounts cannot land before anyone has seen it.
03
Published, and open to reports
After the hold the review publishes on its own. Any Crew member can report it, and three reports from accounts at least three days old hide it straight away.
04
A person decides the exceptions
Hidden and reported reviews go to our exceptions desk, where a staff member signed in with two-factor authentication restores or removes each one, and the decision is logged. A restore sticks: the same reports cannot hide it again.

What gets a review rejected

No substantive cons section, after we ask once. A review with only one side is not publishable here.
The reviewer cannot demonstrate they used the product.
An undisclosed relationship with the vendor, employee, investor, reseller, agency, family.
Signals consistent with machine-generated text, checked by a human before any rejection.
A duplicate of an existing review, including the same experience re-posted under a second account.
Content that identifies a named individual at the vendor, or anything defamatory rather than evaluative.

Reviews are not open yet; the first report is published after the first full quarter. full breakdown

What never gets a review rejected

Being negative. Not once, not ever.
A vendor asking us to remove it.
A vendor paying us anything at all.
Being inconvenient for a product we scored well.
Disagreeing with our own Crash Test result.

Reviews are sorted most recent first by default and never by rating. Suppressing or reordering negative reviews is prohibited by the US FTC rule in force since October 2024 and by equivalent EU and UK consumer law, and it is the specific thing this section exists to rule out.

Provenance

Reviews that did not start here

Starting with empty product pages is the hardest problem a new review site has. Scraping someone else's reviews would solve it, breach their terms, and destroy the independence we are trying to sell. These are the routes that do not.

Written here

×1.00

Native. The reviewer answered our structured questions directly. Labelled as such, because if only imported reviews carried a source line then the absence of a line would stop meaning anything.

Brought across by the reviewer

×1.00

A reviewer who posted elsewhere signs in here, verifies, and answers our questions again in their own words. This produces a native review rather than a republished one. The reviewer owns their experience even where the other platform owns its copy of the text.

Imported by the vendor

×0.55

From in-product surveys or customer interviews the vendor has the right to share. We require the original source, the date and a consent record, and we contact a sample of reviewers ourselves. Labelled with the vendor's name and the source on every card.

Syndicated under licence

×0.40

Licensed from another platform, displayed with the attribution their licence requires and unmodified. A bridge, never a foundation: weighted lowest, phased out as native volume grows, and never marked up as our own review data.

What we will not do

We do not scrape and republish review text from other platforms. We do not display another platform’s star average next to their logo without permission. We do not mark up imported or syndicated reviews as our own review or rating structured data, because those ratings were not collected by us. Deduplication runs across every source so one person’s experience is never counted twice, and we honour takedown requests from original authors and licensors promptly.

Accountability

If we get it wrong

Reviewer appeals

Every rejection states its reason and carries an appeal link. Appeals are read by a different moderator from the one who rejected it, within five working days. If we were wrong, the review is published with its original submission date so it does not lose recency weight for our mistake.

Vendor disputes

A vendor can dispute a factual claim in a review with evidence, and we will investigate and correct or annotate it. A vendor cannot request removal, cannot see who wrote it, and cannot make the process move faster by paying us. Vendors also get exactly one public, labelled response per review which is the whole of their power here, and is genuinely useful when it is specific.

Incentives, and the rule that makes them safe

Incentives are allowed, disclosed on the review card, and never conditioned on what a review says. A one-star review earns exactly what a five-star review earns. The same rule governs gamification: quest and season rewards are for completeness and honesty, never for sentiment, and points can never influence a review score. This is both the FTC’s requirement and the only version that makes the data worth anything.

This page describes our policy, not legal advice. Anyone building something similar should take advice for each market they operate in.