Get Scam Check combines public page observations, fixed evidence weights, and deterministic rule points. It reports what completed, what did not, and why the result has limits.
The checker is intentionally narrow. It looks for defined signals on a limited public surface rather than claiming to know whether a business is legitimate.
The checker accepts a public HTTP or HTTPS store address. It removes the query string and fragment before analysis and storage, parses the registrable domain, and rejects local, private, credential-bearing, or unsupported destinations.
The checker requests the submitted page and follows no more than three validated redirects. It does not sign in, add products to a cart, place an order, or intentionally inspect private account content.
Each configured check is recorded as completed, failed, or unavailable. A completed result can be positive, negative, or neutral. Missing provider data stays visibly unavailable instead of being treated as reassuring.
Versioned rules add, subtract, or leave unchanged a point total. The same structured observations and engine version produce the same points. Most rules cannot apply unless their underlying evidence check completed.
The numeric band is shown only when page access is reliable, weighted coverage reaches the minimum, and no material conflict requires caution. Otherwise the report says Insufficient Evidence, regardless of the point total.
The weights total 100. They measure the ability to support a conclusion, not whether a store is good or bad. A completed neutral check still contributes coverage.
URL validity, domain extraction, HTTPS, public DNS, and redirect chain
28/100
Page access, page title, and meta description
21/100
Contact links and details, refund, about, privacy, and social links
24/100
Aggressive discount and urgency wording
8/100
Domain-ending and brand-impersonation patterns
11/100
Domain age and configured threat intelligence
8/100
Group bars are scaled to the largest group for comparison. A real report lists each individual completed, failed, and unavailable check.
Applied points are added, the HTTPS reduction is subtracted, and the result is capped between 0 and 100. Multiple rules can apply to one report.
Swipe or scroll sideways to read the full rule table.
| Rule | Points | When it can apply |
|---|---|---|
| Baseline uncertainty | +10 | Always applied |
| No HTTPS | +15 | When the submitted address is not encrypted |
| Uses HTTPS | -4 | A small transport-security reduction; never proof |
| Page could not be accessed reliably | +25 | A failed access check also forces insufficient evidence |
| Contact information not found | +8 | Only after the contact checks complete |
| Refund policy link not found | +10 | Only after the page check completes |
| Privacy policy link not found | +7 | Only after the page check completes |
| Aggressive discount language | +10 | When a configured pattern is detected |
| Pressure or urgency language | +10 | When a configured pattern is detected |
| Recently registered domain | +18 | Only with completed age evidence |
| Higher-risk domain ending | +8 | A weak clue from a configured caution list |
| Possible brand impersonation | +18 | When a configured domain pattern matches |
| Threat-intelligence match | +50 | Only when a configured provider reports a match |
| Redirect to another domain | +18 | When the validated chain leaves the submitted site |
| HTTPS downgrade | +20 | When a redirect changes from HTTPS to HTTP |
| Same-site origin change | +3 | A host change that remains on the same site and scheme |
| Two redirects | +4 | A small point increase for a two-step chain |
| Complex redirect chain | +8 | Three redirects or a chain that exceeds the limit |
Visible contact details, A visible refund or return link, A visible privacy link, An about-page link, External social links are recorded for review but do not reduce the total. They are self-published or easily fabricated, so their presence does not verify a seller.
0–30
Still not proof of safety.
31–65
Pause and verify the seller.
66–100
Strong reasons not to proceed yet.
This gate applies even when the arithmetic total falls inside one of the three numeric bands.
Engine version 1.2.0 sets the points, evidence policy, and machine-readable verdict. The version is saved with a real report so its decision can be understood later.
When enabled, a language model may draft plain-language wording from structured signals. It cannot fetch new facts, change the rule points, or change the deterministic verdict.
A wording failure falls back to rule-based copy. Failed and unavailable checks remain in the report instead of being hidden behind a confident summary.
Submit only a public storefront page without personal or access information. Query strings and fragments are removed before analysis and storage. Submitted addresses and resulting reports may be retained to provide the service.
A configured threat provider may receive the normalized public URL and path. If AI wording is enabled, structured signals may be sent to the configured model provider; fetched page HTML is not intentionally sent for that purpose.
Read the complete privacy policyA report describes the public page and available providers at its generation time. It is not continuous monitoring. The page, redirect chain, operator details, promotions, and provider data can change afterward.
Real reports retain their generation time and engine version. Recheck important details before a later purchase, especially if the domain, offer, checkout instructions, or contact information changed.
Explore the synthetic reportApply the method to a real public page