How a second-region re-check stops false uptime alerts
Fewer false alarms come from confirming a failed check in another region before you alert. How that re-check works on a one-region plan, and what it refuses to call an outage.
Most 3am alerts that turn out to be nothing are a bad path between the probe and the target, not a down site. The visitor in the next city is loading the page while you are getting dressed. A second-region re-check exists so the alert waits until two places agree. Fewer false alarms is the whole point; catching a one-second blip is not.
What happens on the first failed check
The monitor is not marked down on that first failure. A confirmation check is queued immediately from a different region — not at the next scheduled interval. If you run every five minutes, you do not sit through another five minutes before anyone looks from elsewhere.
- The confirmation fails too. That single failure is enough. An incident opens without waiting for a second full interval of the original region.
- The confirmation succeeds. The original failure is treated as a local path problem. You are not alerted.
- The original region recovers before the confirmation returns. The monitor stays up. A late failure from the second region does not rewind that.
While the confirmation is in flight, a one-region monitor is held at degraded rather than down. Degraded means “we saw a failure and we are asking someone else”. It is not yet an incident. The default confirmation threshold is two consecutive failures in a region, and it can be set from 1 to 10. The cross-region check is decisive on its own: it does not wait to accumulate that count.
Fewer false alarms on a plan with one region
Free includes one regular region per monitor. That region is chosen by measured latency when the monitor is set to automatic, which is the default for a new monitor: every region is sampled, then the fastest one your plan allows is kept. Solo keeps two, Team and Enterprise keep three. Today those regions are Zurich, New York and Singapore.
The confirmation region is not a second regular region, and it does not use a paid slot. A one-region monitor still gets the re-check. Calling the free plan “multi-region checks” would be wrong: the scheduled checks run from one place. The second place is asked when the first place has already failed.
What this will not filter out
A real outage fails from both regions, and you should hear about it. A site that returns 200 with an error page fails neither check. A region that cannot route to you because you blocked its addresses will look like an outage from that region and, if you blocked both, like an outage from both. Confirmation removes the case where only one vantage point is wrong. It does not remove a target that is wrong everywhere, and it does not make a five-minute schedule notice a 20-second failure that recovered before either check ran.
If you are comparing products because false alarms are why you want to leave, the Pingdom comparison is the page that puts the products next to each other. This article is only the mechanism.