Why uptime monitors send false alarms
A false alarm is an email saying a website is down when visitors can reach it. After a few of them, people stop opening the emails, and then they miss the real one. That is why we treat a false alarm as worse than a slow one.
The usual causes
- The checker's own network had a problem. If the place doing the checking loses its route to part of the internet, every site behind that route looks down from there, while everyone else reaches them fine.
- One slow or failed request. A single timeout can be a busy moment, not an outage. A monitor that alerts on one failed check alerts on every blip.
- A restart or an update. A server restart or a plugin update can make a site unavailable for a few seconds. By the time you open the email, it is back.
- A firewall blocking the checker. Some firewalls and security plugins block visitors they do not recognise. The site is up for people, down for the monitor.
- The monitor itself is unwell. If the checking service is overloaded or broken, it can mistake its own failure for yours, sometimes for every site at once.
What we do before we email you
Each website is requested once every 60 seconds, alternating between two locations on two different companies' networks: Chicago and New Jersey. When a check fails, we do not email yet. We ask the other location to check straight away. Then the first location tries again, 10 seconds after that.
We call a website Down only when both locations and the retry agree within 45 seconds, and only while both locations are known to be working. Then each person on that website's alert list gets one email. When the site comes back, both locations must see it working before each of them gets one recovery email.
How that answers each cause
- A problem on one location's network does not fail the other, because they are on different networks run by different companies.
- One failed request is never enough. It takes three.
- A restart that is over within a few seconds is usually over by the retry, so it does not send an email.
- If one location says down and the other says up, the website shows "Checking" rather than Down. If they still disagree after five minutes, we send one email saying we are not sure. It does not say your site is down, and a second one follows when we can check reliably again.
- If one of our own locations stops answering, its results stop counting, and your website is never called Down on its word. We show that time as a gap in monitoring or as Status uncertain, never as your site being down.
One real day
In the 24 hours to 28 September 2026, the website on our live demo failed a routine check 12 times, each time by taking longer than ten seconds to answer. Four times the other location got an answer straight away. The other eight times both locations waited ten seconds for nothing, and the site answered on the retry, seconds later.
A monitor that alerts on one failed check would have sent 12 emails that day. Asking a second location first would still have sent 8. We sent none, because the site never stayed down long enough to fail all three.
What it costs
Confirming takes a little longer than trusting one check, up to 45 seconds. An outage that is over in a few seconds may never send an email. We think that is the right trade for people who look after other people's websites: an email from us should mean somebody needs to act.
If a firewall blocks us
If it blocks both of our locations, we would say the site is down, and we would be wrong. If it blocks only one, the site shows Status uncertain after five minutes, with one email saying our two locations disagree and one more when both can see it again, never an outage alert. Our two addresses and exactly what each check sends are on about our checks, so you can allow them.
See it before you sign up
Watch a real website being monitored right now, no sign-up, or see the exact email you get when a site goes down. 3 websites are free, forever, with no card. Up to 20 is $12 a month.