How daily rotating visitor IDs work
A daily rotating visitor ID recognizes repeated requests within a UTC day without keeping a persistent anonymous identifier in the browser. It estimates daily activity; it cannot establish lifetime unique people.
From request to anonymous ID
The browser sends a pageview or named event using a project’s public SDK write key. The collector resolves that key to the registered project and site, then combines the UTC date, project, site, request IP, and user agent in a keyed hash. The event pipeline stores the resulting anonymous ID rather than forwarding a raw IP into analytics storage.
Site and project scoping matter: the same request on two sites should not become a universal cross-site identifier. A secret key protects the derivation. Proxy trust also matters because the collector must use the intended client request address rather than blindly trusting any supplied header.
A worked example
Suppose one visitor opens three pages on Monday and returns for two pages on Tuesday with the same request characteristics. Monday has one daily ID and three pageviews; Tuesday has a different daily ID and two pageviews. Across both days, there are two observed daily IDs and five pageviews. It would be incorrect to label those two IDs as two distinct people.
| UTC day | Pageviews | Daily anonymous IDs |
|---|---|---|
| Monday | 3 | 1 |
| Tuesday | 2 | 1 |
| Both days | 5 | 2 distinct daily IDs, not 2 known people |
Sources of counting error
Two people sharing an IP address and user-agent description can collide. A person who changes networks or browsers can produce multiple IDs. Ad blockers, failed requests, late delivery, duplicate installations, and automated traffic also change the observed set. A hash does not make recognition perfect.
The event envelope currently carries this daily identifier as both anonymous_id and session_id. Inactivity-based visit reports reconstruct sessions from timestamps; session_id is not a browser’s 30-minute session token.
What changes when an account logs in
An application can explicitly call identify with its stable account ID. Later events then carry user_id in addition to rotating anonymous context. That can associate identified activity across days without changing the anonymous counting rule. It does not automatically identify everyone who previously shared the daily hash.
Review infrastructure logs, queue payloads, custom properties, and replay separately. A cookie-free SDK describes one part of collection, not every store or legal obligation in your deployment.
Frequently asked questions
Does a daily ID count a returning visitor twice?
A person returning on another UTC day creates another daily anonymous ID. The report measures observed daily identities, not unique lifetime people.
Is the raw IP stored in analytics events?
The collector uses the request IP to derive identity, then analytics storage receives the derived ID. Infrastructure logs must be reviewed separately.
Can I measure cross-day account activity?
Send an explicit stable user ID after login. Anonymous daily identities alone cannot provide that linkage.