What each number means.

1. Collection

The lightweight tracker sends an initial pageview and pageviews for History API navigation. Hash-route tracking is off unless a site explicitly enables it. Custom events can include up to 20 small string, number, boolean, or null properties.

The payload contains the registered site and public collector key, event name, normalized path, viewport width, referrer origin, and only utm_source, utm_medium, utm_campaign, and utm_content. For outbound-link events, only the destination hostname is included. Arbitrary query values, fragments by default, full referrer paths, full outbound URLs, cookies, and local-storage identifiers are not sent.

The Flintglade Analytics acquisition pages use the same four-label campaign allowlist. They preserve those labels through internal navigation and signup and record canonical funnel events server-side without adding an acquisition cookie or browser-storage identifier.

2. Accepted events and plan usage

An event is accepted only after its site and collector key match, the browser origin is compatible with the site, the payload is valid, bot and request controls pass, the workspace can collect, and monthly and independent safety capacity remain. Solo counts accepted events pooled across its three-site workspace. The allowance is 100,000 per UTC calendar month.

Requests that are malformed, unauthorized, recognized as bots, over limits, or received while the workspace is read-only are deliberately not represented in reports. Analytics therefore measures accepted traffic, not every request that may have reached an edge or origin.

3. Visitors

At ingest, a visitor identifier is derived from a secret daily salt, the registered site, request IP address, and user agent. The stored identifier is a truncated one-way hash; the raw IP and full user agent are not stored in analytics event rows. The salt rotates at each UTC day, so the application cannot join one visitor across days.

Visitor counts for a multi-day range are the sum of daily unique visitors, not unique people across the whole range. Shared networks, changing IP addresses or user agents, privacy relays, and multiple devices can merge or split apparent visitors.

4. Visits, bounce, and duration

Events for the same site's daily visitor within a 30-minute inactivity window belong to one visit. A new visit starts after a gap of 30 minutes or more. A bounce is a visit with at most one pageview. Visit duration is the elapsed time between its first and last event, so single-event visits have zero measured duration. Average duration divides total visit duration by visits.

5. Sources and dimensions

The service classifies a referrer's hostname into display source and channel categories. Explicit UTM tags can supply or override campaign context. User-agent and viewport signals produce coarse browser, operating-system, and device categories. A local IP-to-country database provides a coarse country code when available. Classifications are estimates and can be missing or wrong.

6. Events and goals

Page goals match exact paths or explicitly configured prefix patterns. Event goals match an event name. Conversion reports compare distinct daily visitor identifiers that met a goal with visitors represented in the selected range. They show observed attribution, not causal lift, incrementality, or proof that a campaign produced the behavior.

7. Retention and aggregation

Raw events and sessions support filtered detail for 180 days. Completed UTC days are rolled into daily site and dimension aggregates. Site-level aggregates remain until workspace deletion, while wider dimension aggregates can expire after 730 days. Workspace deletion has a 30-day window before permanent purge. See the Privacy notice for account and provider data.

8. Interpretation limits

Ad blockers, disabled JavaScript, network failures, browser lifecycle behavior, bot-classification errors, quota or safety drops, and tracker misconfiguration can reduce counts. Daily identifier rotation prevents cross-day person-level deduplication and funnels. Flintglade is directional operational analytics, not an audited financial, advertising, identity, or compliance measurement system.