A team admin sets a spend alert at a low threshold and expects an early warning email, yet nothing arrives while the month's included usage is steadily consumed. What decides when that alert fires?

The principle — Cursor treats two kinds of consumption separately. Usage included in a plan is already covered by the subscription, while on-demand usage is money spent above it. A spend alert is a money threshold, so it watches the second and ignores the first.

Why the key is correct — The alert stays silent for as long as the team is drawing on included usage, because none of that consumption counts toward the threshold. The first email can only arrive once included usage is exhausted, on-demand usage is enabled, and billable charges have begun to accumulate. Silence therefore tells you nothing about how much of the plan has been used; it tells you only that no on-demand money has been spent yet.

Why the others are wrong — Reading the threshold as a total-consumption meter is the trap this design sets, because it leaves an admin feeling warned while nothing is being measured. A staged alert that fired part-way through the included budget and then again on charges would give that early signal, but there is no such split. An end-of-cycle summary would also arrive far too late to act on, whereas alerts are sent as the threshold is crossed during the cycle. For visibility into included usage, read the dashboard; to cap money rather than be told about it, set a spend limit, which stops AI features instead of sending mail.

Remember this — An alert threshold measures spend, not consumption. A quiet inbox can still mean an empty budget.

Sources — Cursor's spend alerts documentation.