get_audit
ChatGPTFetch a completed CRO audit (or the progress of one still running) by audit id or job id. Returns per-device bucket scores and the failing checks behind them.
get_diagnosis
ChatGPTFull narrative for one monitoring diagnosis: what's happening, the primary cause, urgency, impact, its recommendations and the signals behind it. States the evidence level (L0 = a metric moved, up to L3 = replays show the cause), the measurements and session replays that bear on it, when it was last seen, and which recommendations have no screenshot or replay evidence.
get_fix_brief
ChatGPTImplementation brief for one recommendation: the issue it addresses, the action, the rationale, the measurements behind the finding (rates, counts, benchmark source, relevant session replays), how strong that evidence is, and what the pipeline attached to the recommendation itself. Screenshot and session-replay links appear only when they exist — many recommendations rest on aggregate funnel data alone, and the brief says so explicitly. Includes what to watch to confirm the change worked; Actionbase does not size tests or measure outcomes. Get recommendation ids from get_diagnosis.
get_funnel_history
ChatGPTDaily conversion-rate history for one funnel, newest first, with the change across the window.
get_funnel_overview
ChatGPTLatest conversion snapshot for every tracked funnel on a site, with the worst drop-off step in each.
get_weekly_summary
ChatGPTThis week's digest for a site: tracking state, headline stats, new issues, top open issues and what was resolved.
list_diagnoses
ChatGPTRanked list of conversion issues found by the monitoring pipeline for a site, with headline stats. By default only issues re-detected in the last 14 days are listed; older ones are stale. Use get_diagnosis for the full narrative of any one row.
run_audit
ChatGPTQueue a conversion-rate audit of a page. Returns a job id and the report URL; poll get_audit with the job id for results (a full audit takes a few minutes).
update_diagnosis_status
ChatGPTMove an issue through its lifecycle: investigating, resolved, dismissed, reopened (open) or regressed. Use resolved once a fix is live and the metric has recovered.
update_recommendation_status
ChatGPTRecord what happened to a recommendation: accepted, implemented, or rejected. Rejections take a structured reason and feed back into future runs.