Daily visitor estimate notice
Daily visitor estimates are not counts of verified people. On visible public pages, the browser sends an estimate request containing only the public page path and no visitor ID. It contains no name, email address, form contents, full referrer URL or URL query parameters. The server uses the connection IP address and a broad browser category in memory to create a keyed deduplication digest limited to this website and one calendar day in Beijing time. The visitor-estimate process and database do not persist raw IP addresses or full User-Agent strings. No visitor cookie or daily visitor identifier is stored in the browser, and visitors are not linked across websites or days.
Digests and their independent daily keys are eligible only for the current and preceding Beijing dates. Expired data is cleaned on new estimate events or authorized daily-report reads. Service or report-read interruptions can delay physical cleanup, but expired digests are not used for deduplication. Daily aggregates cover the most recent 90 days; older aggregates are cleaned on new events or report reads.
Use “Do not include my visits” in the footer to stop future estimate events, or “Include my visits” to participate again. The browser stores only a non-unique opt-out boolean, not a visitor identifier. If this preference cannot be saved, estimate events are not sent. Opting out does not retroactively remove existing daily aggregates. Browsers with Do Not Track or Global Privacy Control enabled do not send new estimate events. Marked tests, monitoring and identified bots are excluded.
Different people using a similar browser on a shared network may be combined. One person changing networks, devices or browsers may be counted again. Unidentified bots, disabled JavaScript, blockers and lost requests can also affect coverage. The metric is labelled “daily visitor estimate”, not exact people, customers or sales, and is not used as a session-conversion denominator. Page views, contact clicks and durably saved non-test enquiries retain their separate definitions. Page views are not used to reconstruct visitor estimates before collection starts or during missing coverage.