How scoring works

Methodology

Listing Detective separates evidence extraction from risk scoring. The AI reads the screenshots; fixed rules calculate the score.

1. Screenshot evidence

The AI extracts the item, asking price, seller rating, account-history clues, payment method, buyer-protection language, urgency, and contradictions. It can also estimate a typical resale price when it can identify the item with enough confidence.

2. Price estimate

The typical price is an AI estimate based on general market knowledge. It is not a live lookup of current sold listings. Low-confidence estimates do not lower risk. Unresolved pricing mainly lowers confidence instead of automatically increasing scam risk.

3. Deterministic scoring

Serious warning signs—especially off-platform payment requests, extreme discounts, very new seller accounts, identity conflicts, or contradictory listing details—add risk quickly. Positive signals such as buyer protection or a strong seller rating only reduce risk modestly.

4. Missing evidence

Missing information lowers confidence rather than automatically raising the risk score. The risk score is driven by supported warning signs.

5. What the score means

• Low: relatively reassuring evidence, but normal marketplace safeguards still apply.
• Moderate: important details remain unverified.
• High: meaningful warning signs need independent verification before paying.
• Extreme: avoid buying as presented unless the major red flags can be resolved with strong evidence.

What Listing Detective does not do

It does not authenticate merchandise, confirm a seller’s identity, search live marketplace sales, guarantee buyer protection, or guarantee that a listing is legitimate.