On-Call Cost Calculator
An on-call rotation costs more than the stipend. This adds up the stipends, the time spent on incidents, and the extra weight of being woken at night, then shows it per week, month, year and per page.
An example is filled in. Change every field to match your team.
See where the on-call money goes
What one year of on-call costs, split up
Fixed stipends are paid whether or not anything breaks. Incident time grows with every page, and night pages cost extra.
What goes into the cost
- Stipends. The flat pay for carrying the pager, owed even on a quiet week.
- Incident time. The hours from page to resolution, across everyone involved.
- The interrupt tax. A 3am page is the hour of work plus a slower next day plus the lost thread of whatever came before. The multiplier represents that.
- Attrition (not in the number). Engineers leave teams with punishing on-call, and replacing one costs months of salary. On a rotation that pages nightly, this is usually the largest cost of all.
How to bring it down
- Fewer pages. Delete alerts that never lead to action, page only on user-facing symptoms, and route the rest to a dashboard or a ticket. This is where most of the savings are.
- Prioritise the fixes. An error budget forces the noisy failure modes onto the roadmap instead of being absorbed by whoever is on call.
- Shorter incidents. Runbooks, one-click rollback and tested failover cut the time per page.
- Protect the night. Follow-the-sun handoffs move the pager to a waking timezone; if that is not possible, make sure after-hours pages are rare and well paid.
Paying for it
A flat stipend per shift is the most common model and the easiest to budget. Paying hourly for actual incident time rewards the team for reducing noise. Per-incident pay can quietly reward the opposite. Time off in lieu is common in Europe, sometimes instead of cash and sometimes on top. Whatever the model, agree it before the rotation starts, not after a bad week.
Frequently asked questions
What should on-call cost?
There is no single figure — it depends on team size, how noisy the alerting is, and how you pay for it. This calculator asks for your own numbers instead of quoting an average, because a quiet rotation with a small stipend and a screaming one that pages nightly are worlds apart.
What is the "interrupt tax"?
A page at 3am is not one hour of cost. It is that hour, plus a groggy, less productive next day, plus the context lost on whatever the engineer was doing. A multiplier of 2 to 3 on the raw incident time is a reasonable way to represent it.
What is the biggest hidden cost of on-call?
Attrition. Engineers leave teams with bad on-call, and replacing one costs months of salary in recruiting and ramp-up. That is not in the weekly number here, but it is often the largest cost of a rotation that pages too much.
How do I actually reduce on-call cost?
Cut the number of pages. Delete or re-tune noisy alerts, page only on user-facing symptoms, and use an error budget so reliability work gets prioritised. Then shrink the time per page with runbooks and automation, and consider follow-the-sun handoffs so nobody carries the night.
Flat stipend, hourly, or per-incident pay?
A flat stipend per shift is the most common and the simplest. Hourly pay for actual incident time rewards fixing the noise. Per-incident pay can accidentally reward leaving alerts noisy. Many places also offer time off in lieu instead of, or on top of, cash.