Product Feed Health Checker

Independent preflight checker — not an official Google or OpenAI approval service

Privacy

Privacy policy

What is parsed locally, what the remote scanner receives only after you consent, what the scanned host can see, what is retained and for how long, and which analytics events exist.

Plain summary: your feed file is parsed in your browser and is not uploaded. Remote requests happen only for the URLs you explicitly submit, after a separate consent step. The site you scan can see that a request came from us. Analytics events never contain product values, identifiers or URLs.

1. The feed file

  • Product files are parsed locally, in your browser. The complete file is not uploaded to our server during the default workflow.
  • No account is required to run the first local diagnostic, so the default path involves no identity at all.
  • You can remove the imported file by resetting the workspace or closing the tab. Nothing about the file survives on our side.
  • If a future feature needs to send rows to a server, it will require separate consent and a separate disclosure. It is not part of the current workflow.

2. Remote scan requests (opt-in, separate consent)

Remote checks are a second, explicit decision — never bundled with the import. When you opt in, the scanner receives only:

  • the product-page or image URLs you selected;
  • a random correlation reference generated for that scan;
  • the scan type;
  • market context only where a comparison needs it.

It does not receive product titles, descriptions, SKUs, GTINs, prices or stock levels unless a later, separately disclosed feature requires them. The feed-versus-page comparison happens in your browser after the evidence comes back.

3. What the scanned site can see

Requesting a URL tells that host a request was made: our scanner’s IP address, an honest User-Agent identifying this product, the requested URL and the time. Their logs may record it, exactly as they would for any visitor. We do not impersonate a browser or a search-engine crawler, and we do not attempt to bypass access controls, paywalls or bot protection.

If a host blocks the request, the finding is that access could not be established — which is a fact about access, not a judgement about your image or page.

4. Logging

  • We do not log the full URL including its query string by default.
  • Where logging is necessary: sanitized host, status code, duration, resource category, error code and job ID.
  • Authentication-like query parameters and URL credentials are redacted before anything is written.
  • Cookies, authorization headers, response bodies, feed cells and complete image contents are never logged.

5. Retention

  • The feed file and normalized product records are not persisted on the server at all.
  • Scan job data and structured evidence expire within 24 hours of the job, and sooner once the job completes where the implementation allows it.
  • Page HTML and image binaries are not kept after metadata extraction.
  • Mapping templates stay in your browser and can be cleared by you.
  • Operational logs follow the shortest retention the infrastructure provider offers for the plan in use.

If saved reports or longer-lived account history are introduced, this section will be revised and disclosed before that feature ships — not after.

6. Analytics

Product usage is measured with a small set of named events, and nothing in them identifies what you imported:

  • feed_import_started, feed_import_parsed, mapping_confirmed, validation_started, validation_completed, remote_scan_opted_in, report_exported, revalidation_completed.
  • They exclude product values, identifiers, GTINs, prices, URLs and any part of the file contents.
  • Referral source is recorded only in aggregate where privacy requirements allow it. We do not store user prompts or personal data from an AI assistant.

7. Account and billing data

  • Sign-in is a six-digit code sent to your email address. We store a hash of the code, not the code itself, and the session is a server-side record addressed by an HttpOnly cookie.
  • Order and subscription records hold what is needed to grant and revoke access: plan key, order reference, timestamps and payment state. Card details never reach this application — checkout is hosted by Waffo Pancake.
  • Payment webhooks are verified by signature and deduplicated by event ID; raw payloads and card data are not stored.
  • You can delete your account, which clears pending sign-in codes and removes personal fields from session and event records. Order history is retained with the account identifier because accounting and dispute handling require it.

8. Third parties involved

  • Waffo Pancake processes payments.
  • An email delivery provider sends the sign-in code.
  • The hosting provider runs the site, the account service and the scanner.
  • The merchant sites you choose to scan receive the request described in section 3; they are not acting on our behalf.

Which providers are used in a given deployment is described in the deployment configuration; no provider receives your feed file in the default workflow.

9. Your choices

  • Use the tool without an account, including the full local diagnostic.
  • Decline remote checks and still get every local finding. Remote evidence is additive, not required.
  • Delete your account at any time; see section 7 for what is retained and why.
  • Ask for a correction or deletion by replying through the contact channel in your account area.

10. What this policy does not claim

It does not claim that nothing leaves your device — remote checks are requests to a server by definition. It does not claim a certification or a jurisdiction we have not established. Where the implementation and this page disagree, the implementation is the defect; the changelog records corrections.

Run a local diagnostic with no account