How to monitor client websites as an agency

To monitor client websites, check each one on a schedule for three things: whether it is up, how fast it answers, and when its security certificate expires. Send alerts to a named owner, keep the results in one place, and review them weekly so problems are found before the client finds them.

Clients assume their website works. When it does not, they notice at the worst moment: a customer rings to say the booking page is broken, or a browser warns that the site is not secure. The agency that finds out second looks careless, whether or not it was responsible for hosting.

Decide what you are responsible for

Start with the contract, not the tooling. For each client, write down whether you host the site, maintain it, or only built it. A retainer that includes maintenance implies monitoring. A one-off build does not, but a client will still call you when it fails. Decide in advance what you will do in each case, and say it in the retainer or statement of work.

The three checks that matter

You can monitor dozens of things. These three catch most of the failures a client would actually feel.

Uptime

Uptime is the share of time a site responds correctly. A check requests the page and records the status code. Anything in the 200 range is healthy; a 500 error, a timeout or no answer is down. A single failed check can be a blip, so confirm with a second attempt before raising an alarm. Expressing uptime over 24 hours and over 30 days gives both a recent picture and a trend.

Response time

Response time is how long the server takes to answer. A site can be technically up and still feel broken if every page takes eight seconds. Record the time on every check. Set a threshold for “slow”, for example a response that is consistently well above that site’s normal, and label it clearly. Compare a site with its own history rather than with other clients’ sites, since a heavy ecommerce store and a five page brochure site have different baselines.

Certificate expiry

A TLS certificate is what lets a browser show the padlock and load the site over HTTPS. Each one has an expiry date. When it lapses, browsers show a full page warning and most visitors leave. Many are renewed automatically, which is exactly why they get forgotten until the renewal quietly fails. Check the expiry date daily and warn at 30 days and again at 7 days.

Set up the monitoring

List every client site, including secondary domains and booking or shop subdomains. Missing domains are the usual gap.
Choose the check interval. Every five to fifteen minutes suits most small business sites. A check that runs weekly misses nearly everything.
Name one owner per site. An alert that goes to a shared inbox is an alert nobody acts on.
Write the alert rules. Down after two failed checks. Slow after several consecutive slow responses. Certificate warning at 30 and 7 days.
Decide the response. Who looks first, who contacts the client, and what the client is told. A two line template saves a panicked half hour.
Keep the history. When a client asks “was the site down last Tuesday?”, you want an answer, not a guess.

Tell the client before they tell you

Proactive contact is the whole value of monitoring. A short note such as “your site was unavailable between 2:10 and 2:35 this morning; the host restarted a server and it is back” turns an incident into evidence of care. If the cause is a certificate or a plugin you manage, say so and say what you changed.

Over a quarter, a one paragraph summary in the status update does the same job quietly: uptime figure, slowest day, anything renewed. Clients on a retainer often value this more than they value the work that prevented the problem, because it is the part they can see.

Run a weekly review

Monitoring catches outages. A weekly look catches drift.

Which sites were down, even briefly, and was the cause fixed or only cleared?
Which sites are getting slower over time?
Which certificates expire in the next 30 days?
Which sites have no checks at all because they were added late?

Fifteen minutes with the list, on a fixed day, is enough.

Common mistakes

Checking only the home page. Check the pages that make money, such as booking, checkout and contact.
Alerting on every blip, so people learn to ignore alerts.
Monitoring from one place only, which cannot tell a site outage from a network problem at your end.
Forgetting domains that were registered in the client’s name, where renewals depend on someone else.
Treating green as proof. A page can return 200 and still show an error message in its content.

Where software helps

Kervol’s website health feature checks each client site on a schedule and records uptime, response time and TLS certificate expiry, and shows site health on the agency dashboard alongside the rest of the client’s work. Anyone can run the free website health check on kervol.com to see the same checks for a single site. Any dedicated uptime monitor does the basic job; the point is that someone owns the result.

Checklist

Every client domain is listed, including subdomains.
Each site is checked at least every fifteen minutes.
Uptime, response time and certificate expiry are all recorded.
Each site has one named owner for alerts.
Alert rules and the client message template are written down.
Certificate warnings go out at 30 and 7 days.
A weekly review is in the diary.
The contract states what is and is not covered.