Merchant compliance monitoring

Turtini actively monitors platform activity against the Acceptable Use Policy. Monitoring has three layers, each with its own trigger and each landing in the shared moderation queue in Admin → Moderation.

1. AUP attestation gate — an owner or admin must accept the current AUP version before Stripe Connect onboarding can start (see "Platform Acceptable Use Policy and attestation").

2. Product text moderation — when a product is created or its name/description is edited, Turtini scans the text against keyword groups that mirror Stripe's restricted-business categories (firearms, adult, cannabis, gambling, counterfeit, etc.). Matches mark the product with moderationStatus "pending" and add a conduct entry to Admin → Moderation with type "text" so an admin can review the match and either approve or remove the listing. Clean products pass through silently and are marked "approved". The scan runs only when name or description actually changes — moderation-status flips and routine updates don't re-scan.

3. Nightly conduct sweep — once a day, for every org with charges enabled, Turtini pulls the last 30 days of Stripe charges and computes dispute and refund rates. Orgs flip into "compliance review required" when:
• Dispute rate exceeds 0.75% (Stripe's own high-risk cutoff), or
• Refund rate exceeds 5%, and
• There are at least 20 charges in the window — small samples aren't statistically meaningful.

When a sweep flags an org, a moderation entry with type "conduct" appears in the Admin → Moderation queue. A compliance snapshot is also saved under the org's record every night even when nothing fires, so the history is auditable.

What happens next:
• Admins triage flagged items in Admin → Moderation. The same review flow used for image and text moderation applies to conduct flags.
• Turtini may request additional information from the org, place the Stripe account under review, or — for material AUP violations — suspend or disconnect the Stripe integration.
• Most flags are informational. A single bad month can push rates over threshold without indicating anything systemic, and an admin review is usually enough to clear it.

If you believe your org was flagged incorrectly, open a support ticket from the Help modal and include the date of the flag.