How to get alerts for slow responses

A site can pass every check and still be too slow to use. A slow alert tells you when response times stay much slower than usual, or above a limit you set. It is separate from outages: no incident opens, and slow time doesn’t count as downtime.

Choose how slow is too slow

1. Create or edit a Website or API, Port or Ping monitor.
2. Under Slow alerts, choose Slower than usual, Fixed limit or Off. Monitors you create in the app start with Slower than usual (through the API, set slowAuto); monitors saved before November 5, 2026 keep the limit they had.
3. For a fixed limit, enter Slow limit in milliseconds, such as 2000. Anything from 50 to 60,000 works.
4. Save the monitor.

Slower than usual

• Uptimecheck learns the monitor’s usual response time: the median of its hourly averages over the last 14 days. Alerts start once it has 2 days of successful checks; until then, the monitor’s page says it is still learning.
• The monitor turns slow when three checks in a row are above about 3 times usual, and at least 400 ms more. For a usual 200 ms, that is 600 ms; for a usual 50 ms, 450 ms.
• It is fast again a step lower, at about 2 times usual and at least 250 ms more, so a monitor close to the limit doesn’t flap.
• The monitor’s page shows the usual time and the limit in force, and the response time chart draws the limit. The alert names the usual time too.

How it works

• After each successful check, the last three successful checks are compared with the limit. When all three are slower, the monitor turns slow (amber) and its channels get a slow alert with the average time.
• When the last two are at or under the limit again, the channels hear that it is fast again.
• Failed checks and the re-checks of failures don’t count.
• Slow alerts go out right away, without escalation delays, to every channel of the monitor except SMS. Webhooks get the events monitor.slow and monitor.fast.
• Status pages that show the monitor mark it Degraded while it is slow.

What is timed

For websites and APIs, the whole request: redirects, DNS, connecting, TLS, the wait for the server and the download. For ports, opening the connection; for ping, the round trip. On the monitor’s page, the response time chart shows the median and p95, and each check is split into DNS, connect, TLS, server and download, so you can see where the time goes.

Choosing a fixed limit

Set it well above the usual, for example twice the p95 of the last week, so one busy minute doesn’t alert. A fixed limit suits a promise you make, such as an API that must answer within a second. Since three checks in a row must be slow, the interval decides how soon you hear: with 1-minute checks, after about three minutes.

Page speed alerts

Response time is how fast the server answers. How fast the page feels in a browser is measured by a site health monitor: once a day, Google Lighthouse loads the start page and up to 4 more, on mobile and desktop. Under Alert on a slower page, choose a mobile score of 90, 75, 50 or 30. The alert goes out once when a page’s mobile score falls below it, not every day it stays there.

More on page speed monitoring, and read about website response time.