Features
Everything reads from the same metric definitions, so a number on a dashboard, in an alert and in an automation is always the same number.
Definitions live in one place, versioned, annotated and readable by anyone. Change one and every dashboard that depends on it says so, before the number moves.
Cut by acquisition path, first-week depth or anything else in your event stream. Curves that sit on top of each other get flagged rather than shipped.
A trigger, a condition and an action. Run it in shadow mode alongside a human for two weeks before you let it send anything.
Alerts that arrive where the team already is, with the query that produced them attached, so nobody reverse engineers the number at 8pm.
200ms
The p95 from event acknowledgement to query result, measured across both EU regions over the trailing ninety days. What happens inside your own client SDK buffer is yours.
How it fits
Ether queries where your data already lives. There is no second copy to keep in sync, no reverse ETL to maintain and no window during which two systems disagree.
Three teams describe what that looked like in practice on the customers page.
Documentation
Every guide ends with a working example you can paste into your own workspace.
Connect a warehouse and publish a first dashboard.
15 minDefine, version and annotate a metric so it survives a handover.
25 minTriggers, conditions, actions, and how shadow mode works.
20 minEndpoints, pagination, rate limits and idempotency keys.
ReferencePayload shapes, retries and signature verification.
ReferenceSAML setup for Okta, Entra ID and Google Workspace.
30 minIntegrations
The twelve below cover most stacks. The rest, and the REST API for anything missing, are in the documentation.
Specification
What each plan does with these limits is on the pricing page.