Bot identification and filtering

Basic filtering discards recognized crawler signatures and common automation clients before collection. The tracker also stops in browsers reporting automation. These checks are always active; they do not detect every bot.

Detection matches known client names such as Googlebot, GPTBot, HeadlessChrome and curl as complete tokens. A device model containing the letters “bot” is not enough to discard a visit. Unrecognized automation can still pass these basic checks.

What appears in reports

Sources → Bots groups labels present in accepted records. Historical labelled records may remain after filtering changes. A missing label is not proof that the visitor is human.

Limits

User agents can be spoofed and some automated traffic resembles ordinary browsers. Seean does not store visitor IP addresses or prove whether a visitor uses a VPN. Traffic that never runs the tracker or sends a collection request cannot appear in the reports.

How many requests are blocked

Settings → Usage (website owner) adds a Blocked bots column per month: the number of collector requests the basic filter rejected across your websites, compared with the events counted that month (Web Vitals samples, rate-limited and unknown-token requests are in neither number), and for the current month the matched signature names in lower case with their counts (for example googlebot 412, gptbot 38, browser-automation 5). These are bounded monthly counters in per-website rows separate from the event counters - one small counter update per rejected request, no stored record per request, at most 48 distinct names per counter row with the rest under other - so they show volume and reasons, not individual visits. A rejection storm beyond the website’s normal collection budget is dropped from the counters (they are “at least” figures), never from the filter. Counting started when this feature shipped; earlier months show 0. Requests with an unknown website token, requests to the demo website, and requests declined for Do Not Track or Global Privacy Control are not counted as bots. The filter itself is unchanged: this is visibility, not a stronger filter, and no false-positive rate has been measured.

Usage and exclusions

Discarded requests do not count toward event usage. Requests matching a bot signature are discarded at the collector and never written, so no new event carries a bot label; Sources → Bots only shows labels present in older history. Use browser, path, hostname, IP and country exclusions to exclude your own new activity. Rules do not rewrite previously collected history.

Ignore your own visits

Open Settings → Exclusions for browser opt-out commands and path, hostname, IP address or country rules. Follow the complete exclusions guide for exact matching behavior and how to resume collection.

Something unclear? Contact support.