Mapford documentation
Mapford shows a business the exact web pages AI assistants read before recommending businesses like theirs — and which of those pages they can realistically get onto. It samples each buying question many times per engine, reports every number with an honest range, and scores every cited page as PLACEABLE, GETTABLE, or LOCKED. Agencies resell the resulting report white-label to their clients.
Where to go
- Getting started — for operators (you run the instance). Local install, first agency → brand → prompt set → run → report → share link.
- Agency guide — for agency users. What you can see, white-label settings, requesting runs, explaining the report to clients.
- Methodology (technical) — for anyone checking our math. Sampling protocol, Wilson intervals, the funnel decomposition, the placeability rubric, what we refuse to report.
- Engine internals — for developers. Repo layout, provider adapters, the run execution model, demo mode, the worker.
- Report API — for developers integrating an agency's own tools. The read-only, scoped-bearer-key
/api/v1endpoints, rate limits, and what they deliberately don't return. - Supabase runbook — for the operator, on return day. Moving from the demo deploy to Supabase + real providers in under an hour.
- FAQ & troubleshooting — for everyone. Build issues, data resets, "engine not configured", password recovery, self-serve reset.
If you want the plain-English version of how measurement works, read How we measure first — the technical companion assumes you have.