Monitoring
Health monitor
What is wrong right now, across every account you can reach, grouped by client and by website.
The health monitor answers one question at the scale of a whole book of business: where is something wrong. It gathers every unresolved signal across every account the viewer can reach and groups them by client and by website, with counts of what is broken, what needs a decision and what is queued.
It has no date picker, and that is a design decision rather than an omission. None of its signals are windowed: a broken split is broken, an experiment past its end date is late, a dead one is dead. A time range over that would imply the page re-derives itself per period when it does not.
It also carries no traffic or revenue columns, which keeps it off the aggregation pipeline entirely. The practical consequence is that the page stays correct even when the numbers behind the reporting screens are catching up. Volume over a period is what the dashboard is for.
What the viewer sees depends on who they are rather than on what they ask for. An agency sees its client roster; a direct customer sees their own account. The scope is resolved from the signed-in session, so there is no account identifier in the address to check and none to tamper with.