Actions and agent analytics

Actions and agent analytics

Turn evidence into owned or earned work, then inspect how AI agents access the site.

Actions turn observed source and visibility gaps into trackable work. They should begin with evidence such as a competitor citation, missing owned page, crawl restriction, or repeated answer pattern. Accepting an action records intent; it does not prove that the underlying opportunity has been completed or produced impact.

Agent analytics covers how AI crawlers interact with the site when compatible external evidence is available. Crawlability checks and server-log analysis answer different questions: one describes whether access appears possible, while the other shows actual request behavior. Treat missing log data as missing evidence, not as zero crawler activity.

An action should have an owner, evidence link, intended change, and a condition for review. Impact cannot be inferred simply because work was marked complete. Compare later scans using a stable population and look for the expected change in answers, citations, crawl behavior, or page evidence. External events and model updates should remain plausible alternative explanations.

Verified against the current productUpdated July 26, 2026
AreaWhat it containsTypical outcome
EarnedOff-page opportunities derived from citations and competitor source gaps.Outreach, inclusion, reviews, partnerships, reference coverage
OwnedOn-page opportunities derived from crawled content and answer gaps.Content, structure, entity clarity, metadata, or technical changes
ImpactAccepted work, status, history, and later evidence.A record of what changed and whether the signal moved
Crawl insightsAI agent visits from a log provider or uploaded access data.Bot, host, path, and response-status analysis
Crawlabilityrobots.txt behavior and per-URL testing across known AI bots.Confirm access policy and identify blocked surfaces

Crawl insights needs external data

The page provides a guided connection flow, but useful traffic analysis starts only after compatible server-log data is connected or uploaded.

Document the connected source, supported format, date coverage, and freshness before interpreting activity. No imported rows means the analysis is unavailable, not that AI crawlers made no requests.