Agency On-Call: Handling Client Outages Without Burning Out

Running client sites means one day you will get the 2 a.m. message: "our site is down, can you check?" How that moment goes decides whether you are a hero or a problem. The goal is not to be faster at panicking — it is to make sure you are woken only when it is real, and that you already know what to do when you are.

Step 1: Make Alerts Mean Something

Most on-call burnout is caused by alerts, not outages. The fix is upstream of the pager:

• Use failure thresholds. Alert after 2–3 consecutive failed checks, not on one slow request. A single dropped packet is not an incident.
• Confirm from more than one place. Multi-region checks mean a regional hiccup does not page you — only a real outage does.
• Monitor the right things. Watch the checkout, the API, the forms — not just "is the homepage 200." See our guide to fixing alert fatigue.
• Schedule around known work. Maintenance windows for deploys and migrations, so planned downtime never looks like an incident.

Step 2: Severity, Not Just "Down"

Not all client outages are equal. Before an incident, agree internally what matters:

• Critical — an e-commerce or SaaS client's core flow is down (sales or signups stop). Page someone now.
• High — a brochure site or marketing page is unreachable. Email on-call; page only if it persists.
• Low — an SSL certificate expires in two weeks, a domain is at risk. Those never need a 3 a.m. page because you caught them early.

That last point is why SSL expiry, domain expiry, and blacklist monitors matter for agencies: the things that become emergencies later are exactly the ones you can catch weeks in advance.

Step 3: Tell the Client Before They Ask

The worst case is not the outage — it is the client discovering it themselves and assuming you were asleep. Two habits remove that:

• A live status page the client can check (and share with their customers). When the client sees it flipping to degraded before you have even emailed, you look proactive, not caught out. See how to build a public status page.
• A short, factual update: what is affected, what you are doing, next check-in. Three sentences. Nobody needs the full postmortem at 2 a.m.

Step 4: Know Who Fixes What

Many agency outages live at the boundary between your work and someone else's — the host, the domain registrar, the payment gateway, a third-party plugin. Document it per client: who to call for hosting, where domains are registered, which services are client-managed. During an incident you should not be guessing where the DNS is hosted.

Step 5: Debrief in the Light of Day

After the outage, write the one-pager: what happened, when, impact, what you changed so it will not happen again, and the uptime number for the period. On Monitorion you can generate that report per client automatically, which turns a bad night into proof that you were on it — and a natural renewal conversation.

Good monitoring makes on-call boring, and boring is the goal. Create a free account and set up thresholds, maintenance windows, and per-client projects today.