The worst way to find out a client's site is down is a text from the client. Monitoring flips that around, so you're the one saying "we already caught it." But "is it up?" is only the first question, and on its own it will happily report that everything is fine while something expensive is happening. Here's what to watch — and what our own site did in September while every uptime check stayed green.
The four things worth watching
1. Uptime, checked from more than one place
Is the site responding? A good monitor checks every minute or two from more than one location, so a single blip doesn't page anyone but a real outage does within minutes.
Where the check runs from matters more than most people expect. For a while this year, the security rules in front of our own domain were refusing requests from cloud datacentre addresses while every ordinary browser got the page. Visitors saw nothing wrong. Behind the scenes, one of our services couldn't reach another, and customers were shown the wrong plan until we traced it. A monitor that only ever checks from one kind of network can be confidently wrong in either direction.
2. SSL certificates, which now expire faster every year
An expired certificate puts a full-page security warning in front of every visitor, and it's entirely avoidable. It's also becoming a more frequent job. The industry body that sets certificate rules has cut the maximum lifetime to 200 days from 15 March 2026, then 100 days from March 2027 and 47 days from March 2029. Renewal has to be automated, and the automation has to be watched: a renewal job that quietly stops working now gives you weeks of warning, not most of a year.
3. DNS and domains
Domains expire. DNS records get changed by someone "just trying something". Either can take a site — or its email — offline in a way that's hard to diagnose under pressure. Watch the records and the renewal dates, so the slow-motion failures get caught while they're still slow.
4. Speed — and the one almost nobody watches: bandwidth
Slow is the new down. A site that takes eight seconds on a phone is losing people while it's technically up, so track load times over weeks, not once.
Then bandwidth. Between 6 and 9 September 2026 our own marketing site ran at a flat 317 MB an hour for 71 straight hours — 22 GB, about 220 times its normal rate. Uptime stayed green the whole time, because the site was up. The cause was almost certainly ours: a verification script left polling a single 647 KB page every few seconds, on a page set up so every request went through to the server instead of being served from cache.
No alert caught it. The shape did. Real visitors rise and fall through the day; this was a perfectly flat line at three in the morning, which is what a machine looks like. Watch request volume and bandwidth, not just the up/down light.
Why it's really about trust
Clients can't see the dozens of things that go right every day. They can see the one that goes wrong. Monitoring turns that invisible reliability into something you can show them — the kind of message you want to be sending:
"Heads up — your SSL certificate was due to renew next week and the automatic renewal had failed, so we've sorted it. Nothing for you to do."
A short monthly summary of uptime and the problems you headed off makes the case for your retainer without you ever having to ask.
Set it up once, then let it do the remembering
Checking certificates and speed by hand never lasts. The point of monitoring is that it runs whether or not anyone remembers, and only interrupts a person when a person is needed. Set sensible thresholds, send alerts somewhere they'll actually be read, and put a status page in front of clients under your own brand.
Want round-the-clock, white-label monitoring with alerts?
See Monitoring plans →