How we verify and moderate
YayNo checks that a review came from a real visit, and that a person who claims a business actually represents it. Machines recommend. People decide. Businesses can never pay to hide a review, and we never take a review down just because it is negative.
How we verify a business claim
- The claimant uploads two documents: a business document that names them (owner, member, officer, or registered agent), and a government photo ID whose name matches.
- We read both documents (OCR and field extraction), fuzzy-match names and addresses against the listing and the account, check expiry dates, and look for editing-software traces.
- We look the business up in public registries (OpenCorporates when a plan is approved; state Secretary of State adapters, starting with Alaska DCCED, read-only and polite) and against our Overture Maps directory. IRS TIN matching is a manual e-Services process — we never auto-submit a TIN.
- We can look up the listed phone's line type and carrier, and check whether the claimant's email sits on the website's domain.
- An optional third-party ID check (Stripe Identity, Persona, Onfido/Entrust, or Veriff), including selfie liveness, runs only if the claimant consents to sending their ID to that vendor. Nothing is connected in this demo.
- Every check becomes an explainable reason on a 0–100 risk score. The AI never auto-approves a claim. A human admin decides. If they approve, a 72-hour hold still runs, and the listed contact can say "this wasn't me."
How we moderate reviews
Every review, photo, shout-out, and business-profile edit is checked in real time for:
- Toxicity, harassment, threats, hate, spam, off-topic promotions, and conflicts of interest.
- Personal information or doxxing.
- Undisclosed incentive language ("got a discount for this review"). Disclosing a reward through YayNo's incentive field is allowed; hiding one is not.
- Photos: nudity, violence, faces of minors, and text in images.
- Fraud signals: device-fingerprint reuse, network velocity, review bursts, geo-impossible visits, near-identical wording across accounts, owner self-reviews, and competitor brigading.
Reviews are never removed or held for being negative. The FTC Consumer Review Fairness Act and the FTC's Consumer Review Rule bar suppression of honest negative opinions. Holds are for policy violations only.
If we hold something, the author sees the reason and can appeal. Every hold and every decision is audit-logged. Moderators can act in bulk, and those actions are logged too.
Privacy
- Government ID images are encrypted at rest, visible only to the admin reviewing the claim, and every view is logged. ID numbers and dates of birth are masked by default.
- ID images are sent to a third-party vendor only with the claimant's explicit consent.
- Government IDs are deleted 30 days after a decision; business documents after 90 days.
- We do not put ID material on public pages or in exports.
Transparency
These counts are aggregates. They include demo and sample data on this preview.
| Held for a policy check | 0 |
|---|---|
| Rejected for a policy violation | 0 |
| Appealed by the author | 0 |
| Holds overturned (published) | 0 |
How visit verification works · Content policy (draft) · Privacy (draft)