Engineering · 4 min read
Why the engineers who build it should be the ones on call
We operate what we ship, and the people who wrote the code carry the pager. It changes how they build, and it is the cheapest reliability investment we know of.
Tomasz Wróbel
November 4, 2026
Every system we deliver comes with a monitoring dashboard, an alerting policy and a rota. The people on that rota are the people who built the system. Not a separate support team, not a junior who inherited it, the engineers who made the design decisions.
This is not a staffing accident. It is the single thing that most changes how software gets built.
What changes when you expect to be paged
An engineer who knows they will be woken up by their own code writes it differently. Specifically:
- Errors become informative. Nobody wants to be woken at three in the morning by an alert that says "something failed". So the alerts say what failed, for which customer, and what to check first.
- Retries and timeouts get real thought. A flaky integration that is somebody else's problem stays flaky. One that pages you gets a retry policy, a circuit breaker and a dashboard within a week.
- Deployments get safer. Rollback becomes a button rather than a plan, because the person pressing it at midnight is the person who built it.
- Scope gets more honest. The feature that would be clever but fragile gets argued against in design review by the person who would carry it.
What it looks like in practice
Each system has a short runbook written by the team, with the three or four failure modes they expect and what to do about each. Alerts page for things that need a human now and go to a channel for things that can wait until morning. We review every page in the following week: was it real, was it actionable, and what would stop it happening again.
Over time the pager gets quieter. That is the point. A system that pages rarely is one that has had its rough edges filed off by the people most motivated to file them.
Why clients care
When something goes wrong at two in the morning, the person who responds is one who knows why the system is built the way it is. There is no ticket queue, no escalation path, no "I'll need to check with the original team". That is also why a question in year two gets answered by someone who remembers the decision from year one.
It costs us something: senior engineers on a rota rather than a cheaper support desk. We think it is the best reliability money anyone spends.