Confiance et sécurité

Comment fonctionne la modération, ce que nous autorisons ou non, et comment nous traitons les abus.

Cette page n'est pas encore traduite. La version anglaise est affichée ci-dessous. Pour aider à la traduire, écrivez à [email protected].

Last updated 12 Sep 2026

Nothing goes live unchecked

A rating on wowpilot is screened before anyone else can see it, never only after a complaint. The first pass is automatic — length, links, shouting, personal information, how many ratings this account and this network have sent today, how old the account is. Anything that trips a check waits for a moderator. Everything else publishes, usually within a business day.

How fakes get caught

  • Limits. One rating per person per business, and only a handful per account per day.
  • Hashed addresses. We store a one-way hash of the sending IP, never the address itself — enough to notice ten "different" customers posting the same paragraph from one connection.
  • Account weight. A rating from a brand-new account with no history counts for less in a score than one from someone with a track record, and repeatedly-flagged accounts lose weight further.
  • Burst detection. A quiet business that suddenly collects a run of five-star ratings has the newest ones held until a person has looked.
  • Bot check. Sign-up, ratings from unverified accounts and anonymous reports all pass through Cloudflare Turnstile.

Criticism isn't for sale

There is no way for a business to pay for the removal of a genuine negative rating — not a plan, not a partner programme, not a phone call. What a business can do is reply, in public, right under the rating. A rating only comes down if it breaks the rating rules, and that decision is made by a moderator who doesn't know whether the business pays us anything.

Both sides on the record

Once a page is claimed (how that's verified), the owner and their team can reply to any rating, and the customer can reply back. Replies are public and labelled, so a reader always sees who said what. We think a good reply to a bad rating is the most persuasive thing on the page — and we'd rather businesses argue their case than bury it.

Reports from the community

Any member can report a rating as spam, fake, abusive, off-topic, containing personal details, or "something else". Three unresolved reports hide a rating automatically until a moderator looks again. Reports are read by a person; using them as a weapon gets the reporting account restricted.

Every decision is logged

Publish, hold, hide, remove, restore — each moderator action is recorded with who did it and why. Removed ratings are excluded from scores and search but kept, so an appeal can reverse a mistake and a pattern of abuse stays investigable.

Who gets to speak for a business

A page is only "claimed" once the owner proves control of the business's domain: a role-address email, a DNS TXT or CNAME record, an HTML file, or a meta tag — all checked by our servers. Tokens expire after 7 days. A claim built on a misrepresentation is revoked when found.

What we can't promise

No filter catches everything. A patient attacker with real accounts and real domains can still get a few fakes through. Our answer is to make that expensive, keep the audit trail complete, and act fast when someone shows us a problem — not to pretend the problem doesn't exist.

Tell us

A rating that shouldn't be up: use Report this rating on the page. A whole account, a campaign, a security hole: write to us and a person will pick it up.