A monitor set is never finished. A site gets retired, a backup moves to a new time, a developer leaves and takes the knowledge of why an alert exists with them. Six months later, a few monitors are watching things that no longer matter, and the alerts they send get ignored. The fix is small: a short, regular tidy-up. Fifteen minutes a month is enough for a team of up to ten people.
Why monitors drift
Most monitor problems come from change, not from mistakes. Three things cause nearly all of it.
Sites are retired. A client moves to a new domain, a campaign page is taken down, or a staging site is no longer needed. The monitor keeps running and keeps reporting a site that nobody uses.
Jobs change schedule. A nightly backup becomes an hourly one, or moves to a weekend. If the heartbeat monitor still expects the old pattern, it either alerts when nothing is wrong or stays quiet when something is.
People leave. The person who set up the alerts may have been the only one who knew why a monitor was there. The people now receiving alerts may not be the right ones.
None of this needs fixing in a crisis. It needs checking on a schedule, before the noise makes people stop paying attention.
Step one: list every monitor
Open your list of monitors and go through it one line at a time. For each one, ask a simple question: does the thing it watches still exist, and does anyone still care if it breaks?
If the answer is no, delete the monitor or pause it. Pausing is useful when you are not sure. Leave it paused for a month, and if nobody asks for it, delete it.
Write down anything you are unsure about. You will come back to it in step two.
Step two: read the last month of alerts
Next, look at what the monitors actually sent you over the last month. Ask one question of each alert: did someone act on it?
An alert that led to a fix is useful. Keep that monitor as it is.
An alert that nobody acted on is a signal. Either the problem was not real, the problem was real but not important, or the alert was not sent to the right person. Note which monitors produced these alerts. Those are the ones to look at in step three.
A steady stream of ignored alerts is the main reason teams stop reading them. Cutting that stream is the point of the whole exercise.
Step three: adjust the failure threshold or check interval for monitors that alerted for nothing
For the monitors that alerted for nothing, you have two settings to review.
The failure threshold sets how many failed checks in a row a monitor needs before it alerts. If a site had one slow response and you were alerted, raising the threshold means a single blip no longer sends a message. Set it to a number that matches how fast a real problem needs your attention. A marketing page that can wait an hour does not need the same threshold as a checkout.
The check interval sets how often the monitor looks. A very frequent check on a page that rarely changes adds noise without adding much value. Lengthening the interval reduces checks, but it also means problems are found later, so only change it where the risk is low.
Make one change at a time, and note what you changed. If the next month is quieter and nothing important was missed, the change worked. If something important was missed, put the setting back.
Step four: check each heartbeat interval
Heartbeat monitors are for cron jobs and backups. The job checks in each time it succeeds, and you get an alert when a check-in is late. That only works if the heartbeat interval matches how often the job really runs.
For each heartbeat monitor, compare two things: the interval you set and the schedule the job runs on today. Look at the job itself, not your memory of it. If the job now runs every six hours, and the monitor expects every twenty-four, you will either be alerted late or never. If the job runs on a weekly schedule, a daily expectation will alert you every day for nothing.
Update the interval wherever the two no longer match. Then leave a short note on the monitor saying what the job is, so the next person does not have to work it out again.
Step five: check alert recipients
Now check who gets the alerts. Alert emails go only to people in your organisation, so the question is whether those people are still the right ones.
Look at each monitor's alert recipients. Remove anyone who has left, changed role, or no longer looks after that site or job. Add anyone who now owns it. For a small team, this often means one name changes each quarter, and it is easy to miss until an alert goes to nobody useful.
If a monitor has no clear owner, that is worth knowing. Either give it one or decide that it should be deleted.
Step six: look at your status page
Your status page is what customers see during an incident, so it has to be accurate. Open it and read it as a customer would.
Remove any component that no longer exists. Check that the names still match what your customers call the service. If a past incident is still shown as open, close it. If an old maintenance notice is still on the page, take it down.
A status page full of old components makes customers doubt the rest of it. Ten minutes of cleaning keeps it trustworthy for the next incident.
Make it a recurring task with one owner
The tidy-up only works if it happens every month. Put it on your task list as a recurring task, and give it one owner. One person keeps the task moving. Others can help, but one name is attached to it, so it does not drift the way the monitors do.
After each tidy-up, write two lines in the task: what changed, and why. For example, "Paused the old staging monitor, site retired. Raised the failure threshold on the blog to three, it alerted for single slow responses." Those notes are what the next owner will need when they wonder why a monitor looks the way it does.
If you are already running monitors on the free plan, you have room for ten of them, which is enough to start. Start free at https://app.aperioflow.co.uk/#/signup, set up the monitors you rely on, and put the monthly tidy-up on your task list today. Our earlier posts on failure thresholds, check intervals and building an alert rota cover the settings in more detail, and are worth reading before you change anything.