GEO API
GEO (Generative Engine Optimization) is Iron’s AI-search visibility tracker. It measures whether
an LLM answer engine names, cites, or ranks a client’s local business for the prompts a real
prospect asks (for example, “best lawn care in Austin”). It replaces external trackers with an
in-house, org-scoped implementation. Endpoints are under /api/v1/geo.
Endpoints
Section titled “Endpoints”| Method | Path | Description |
|---|---|---|
GET |
/geo/config |
Get the org’s tracked-prompt config (one active config per org; null if unset). |
PUT |
/geo/config |
Set the tracked-prompt config. |
POST |
/geo/config/seed-from-gbp |
Seed prompts from the client’s Google Business Profile. |
POST |
/geo/run |
Run a visibility check now (409 if no config). |
GET |
/geo/overview |
Latest run per engine — the dashboard snapshot. |
GET |
/geo/runs |
Run history (time series). |
GET |
/geo/runs/{run_id} |
One run with per-prompt detail. |
GET |
/geo/checklist |
The static Core-30 / answer-engine optimization rubric. |
How scoring works
Section titled “How scoring works”For each tracked prompt the scorer measures brand presence, ordinal position, citation, and
competitor share-of-voice from an answer engine’s response. The neutral programmatic engine is
Tavily; captured consumer-engine answers can also be submitted via POST /geo/run. A run that
scores zero answers writes nothing, so an empty run never corrupts the time series.
A monthly beat runs the check for every org with an active config (monthly because the free Tavily key has a daily cap and AI recrawl lags days to weeks). The beat is a no-op without a Tavily key configured.
The on-page GEO layer (place-id JSON-LD, freshness/HowTo schema, AI-bot robots rules,
llms.txt, combo pages) ships in the funnels’ shared Astro templates; this module surfaces the
checklist and measures the result.