Visit Intelligence
What the Visit Intelligence console shows about AI-agent traffic on your site — live feed, funnel, consideration, benchmarks, journeys and traces — how fresh the figures are, and how relabelling past traffic works.
Visit Intelligence shows the AI agents visiting your property: who they are, how well they are verified, what they look at, and what they buy. Open it from Visit Intelligence in the dashboard. Most screens share the same filters: Period (today, last 7, 30 or 90 days), Platform, and on the live feed Checkpoint (the stage of a purchase a request reached).
The screens
| Screen | What it shows |
|---|---|
| Live feed | Verified inbound agent activity as it happens, plus Verified vs claimed: how much traffic is verified, identified but unregistered, declared but unverified, or anonymous. Generate audit report opens a print-ready Agent Traffic Audit of the last 30 days. |
| Funnel | The agent conversion funnel — how many agents move through each stage — overall and by platform. |
| Journeys | Identify-later: visitors first seen anonymously and identified later, by platform and by first-touch path; token-linked journeys; and where traffic came from. |
| Consideration | Consideration & lost offers: what agents evaluated versus bought, by platform, and the evaluations that didn't turn into a purchase. |
| Traces | One visit end to end: its timeline and the policy decision behind each step. |
| Benchmarks | Agent commerce benchmarks for platform-attributable traffic, including how machine-readable your site is to agents. |
Identify-later answers the window you choose — up to 365 days. If a longer window isn't available at that moment, the screen says so ("The identify-later figures cover the last N days, not M") rather than quietly showing a shorter one.
Where your agent traffic ended up
The Where your agent traffic ended up panel splits agent traffic into three outcomes:
- Passed: reached your site and was not turned away.
- Denied after arrival: AstraSync's policy turned it away, either at your edge adapter in enforce mode or at your origin. The live feed shows the reason, for example "Blocked by the AstraSync edge · not registered".
- Blocked at your edge: your own CDN or WAF blocked or challenged it. Your edge's firewall and bot rules run before any AstraSync code, so this traffic only appears once you connect your edge logs. Without them, the panel and the audit report say not measured, never zero.
Below the bands, Missed opportunities lists the agents your edge blocked by platform, and the rules that blocked them. Requests that presented agent credentials (Web Bot Auth or AstraSync) are counted separately. Those agents were identifying themselves.
Blocked-at-edge rows are classified, not verified. They come from your logs, and a blocked request never reached AstraSync to prove who sent it. Platform names come from fingerprints and from the bot names your own WAF assigned.
How we recognise agents
The user-agent string is only one signal, and many agents don't send a distinctive one. AstraSync attributes traffic using any of:
- Credentials: an AstraSync identity, or a Web Bot Auth signature (its
Signature-Agentnames the agent's operator even when the user-agent looks like Chrome). - Request headers that platform SDKs and agentic browsers add, including
sec-ch-uabrands. - TLS fingerprint (JA4) and network (ASN), where your edge provides them.
- Published IP ranges for vendors that publish where their agents connect from.
- Your WAF's own bot names, for requests read from your edge logs.
Unattributed automated traffic with a stable fingerprint is grouped and given an honest label until it can be named. When AstraSync learns a new agent, recent history is relabelled (see below).
Some agents drive a real browser from residential connections, don't sign their requests, and publish no ranges. Nothing on the wire distinguishes them from a person, for AstraSync or for your WAF, and they count as human traffic. Agents become visible when they present an identity: AstraSync registration or Web Bot Auth.
How fresh the figures are
Most figures come from summaries that AstraSync keeps up to date continuously, bringing the newest hours up to date first. Each screen shows an As of time — the point the figures were current at. A screen marked Live figures is computed directly from your traffic at the moment you open it.
"Couldn't load" and "took too long"
A screen that fails to load says so — it never shows zeros or "no data yet" in place of an error:
- "Couldn't load … — Retry": the request failed. Press Retry.
- "… took too long to load — Retry": the request timed out. Press Retry.
If figures were already on screen, they stay there with "Showing the last figures loaded" until the retry succeeds. "No agent traffic yet" (or "No funnel data yet", and so on) appears only when the data really is empty for the chosen window.
Relabelling past traffic
AstraSync keeps learning to recognise agent platforms. When it starts recognising a platform it missed before, your earlier visits from that platform — previously counted as unknown — can be relabelled so your history reads the same way as today.
Each account chooses, under Relabelling past traffic on the Live feed screen:
- Relabel past traffic on or off. Off keeps past figures exactly as they were first shown.
- How far back — 1 to 400 days (90 by default).
Until you save your own choice, your account uses AstraSync's default for all accounts (relabel on, 90 days).
Visit Intelligence reads the traffic your edge adapter or Verification Gateway reports, plus your edge logs for requests your CDN or WAF blocked.