Pick your critical journeys
Choose the homepage, a key service page and the feature your business relies on for inquiries or orders (lead generation). For each check, write down the expected result. Don't stop at checking a server port if what you need to monitor is the application.
Functional tests shouldn't create real payments or emails without approval. Use a dedicated test mode or a check that doesn't send anything outside.
Track slowdowns, not just full outages
Repeatedly slow responses can hurt visitors even without an error. Keep a history of measurements and test conditions. One random slow request shouldn't automatically set off an alarm for the whole team.
Separately keep an eye on certificate expiry and critical scheduled tasks. For a blog, that might be a post going live; for a store, a data update. A green homepage tells you nothing about these processes.
An alert has to lead to action
The alert should include the URL, the symptom, the time and how often it repeats. Name a primary and a backup person responsible. After recovery, record the cause and retest the exact scenario that broke.
Monitoring is only useful when every alert has a person behind it who knows what to do. It's part of website maintenance. For server solutions , it's better to separate server monitoring from site monitoring, and if the problem is speed, see our breakdown of why your website is slow.
