Product analytics is coming soon and is available on plans with product analytics. Until it
launches, the Web analytics and Product pages show Coming soon, and only projects with early
access can send or read product events.
What you get
- Web analytics: visitors, sessions, page views, pages per session, bounce rate, and average session length, with the change against the previous period, a traffic chart with your releases marked on it, and breakdown cards for pages (top, entry, and exit), sources (channels and referrers), campaigns (UTM source, medium, and campaign), countries, and devices (device type, browser, and operating system). View all opens the full table of a card.
- Filters: select any value in a card or table to filter the page by it. Filters show as chips that you can remove, and they stay in the page URL, so you can share a filtered view.
- Live visitors: the number of people with an event in the last five minutes, refreshed every 30 seconds while the page is open.
- Goals: count conversions for an event (optionally with one property value, for example
plan = pro) or for a page path such as/checkout/successor/blog/*. The overview shows the conversions and the conversion rate of each goal for the period and the filters. - Check installation: shows whether product events arrived in the last 24 hours, from which page hosts and environments, and the setup snippet with the consent rule.
- AI traffic: requests from known AI crawlers and AI agents to your server, from your traces, with a confidence label. A user agent alone is a claim, not proof.
- Product events: every custom event name with counts and visitors, and the newest events. Select an event to see its property keys with their top values; select a value to filter by it.
- Funnels: ordered steps (page views and events) inside a time window, with the median and 90th percentile time between steps. For each drop, the error groups in the sessions of the people who left, with links to the issue and to replays.
- Privacy controls: export or delete all product events and identity mappings for one distinct or account ID from the project-scoped API.
Privacy first
- Consent: the SDK sends no product events until you call
setConsent("granted"). Before that it drops events (or, withconsentBuffer: "pending", keeps at most 50 in memory), and it stores nothing in the browser. Each event carriesconsent: "granted", and Squasher rejects events without it. You are responsible for getting consent from your visitors. - Scrubbing: the SDK removes query strings and fragments and masks ID-like path segments (IDs, tokens, email addresses) before it sends a page view. Only query parameters that you allow are sent, as event properties.
- No IP addresses: Squasher stores a country code, not the IP address. It also records the host of the page that sent the events (from the browser request), so the installation check can show where events come from.
- Limits: at most 32 properties per event, 64-character keys, and 512-character values.
- Delete local data:
reset()starts a new anonymous visitor at logout;forget()removes the stored IDs and turns consent off.
Filters
A filter is a dimension and a value, for examplepage:/pricing, country:DE, or prop.plan:pro.
Values of the same dimension are alternatives; filters on different dimensions must all match.
Entry and exit page values filter by page. You can apply up to 10 filters.
Definitions
- Session: one visit in one browser tab, the events with the same session ID.
- Bounce rate: the share of sessions with exactly one page view.
- Pages per session: page views divided by sessions.
- Average session: the time from the first to the last event of a session. A session with one event counts as 0 seconds.
- Channel: from the first event of a session. Paid and email come from
utm_medium; search and social from the referrer; other referrers are referral; a session with no referrer and no UTM source is direct. - Entry and exit pages: the first and the last page view of a session. Their second column counts sessions, not page views.
- Goal conversions: people who completed the goal at least once in the period. The conversion rate divides them by all visitors in the period with the same filters.
Counting
Delivery can repeat an event, so every event has an ID and counts once. Visitors are approximate unique anonymous IDs. A visitor who callsidentify counts as one person with every other visitor
that has the same user ID.
API, CLI, and MCP
Use the Product analytics API,squasher analytics in the CLI, or
search("analytics") in MCP Code Mode. API keys need the analytics:read scope to read and
analytics:write to manage saved funnels and goals.