On-Call Rotation for Small Teams: Schedules That Work

On-call rotation for small teams cover: on-call without burnout, with the PagerDuty logo on a navy background

A small team can run a fair on-call rotation by keeping one primary and one backup, rotating weekly once there are three or more people, paging only for urgent problems, and paying or giving time off for out-of-hours cover. Full 24/7 cover needs more people than most small teams have: Google puts the single-site minimum at eight.

Key takeaways

  • Google's SRE book derives a minimum of eight engineers for a single-site 24/7 rotation, assuming a primary and a secondary at all times, week-long shifts, and no more than 25% of time on call.
  • For two people, Atlassian suggests alternating days; for three or more, weekly rotations. Always name a backup.
  • Follow-the-sun means each site is on call only during its own daytime. It needs people in different time zones and removes night shifts.
  • Google's ceiling is two incidents per 12-hour shift. If your rotation regularly exceeds that, fix the alerts or the system before you add people.
  • PagerDuty's pricing page on 13 November 2026: Free for up to 5 users, Professional at $21 per user a month billed annually or $25 monthly.
  • Our estimate for an in-house 24/7 team of four mid-senior engineers is $832,000 a year in the US or £441,800 in the UK.

This guide is for small engineering teams that already get paged, or are about to promise customers 24/7 support, and need a schedule that people can sustain. Figures from Google, Atlassian and PagerDuty are linked to their sources; the cost model is our own and its assumptions are listed.

How many engineers do you need for an on-call rotation?

For full 24/7 cover with two people always on call, more than most small teams have. Google's SRE book starts from its rule that no more than 25% of an SRE's time is spent on call. From that it derives that "assuming that there are always two people on-call (primary and secondary, with different duties), the minimum number of engineers needed for on-call duty from a single-site team is eight", with week-long shifts, so each engineer is on call for one week every month. For dual-site teams it suggests at least six per site (Being On-Call).

Few startups have eight people who can debug production, so in practice small teams make one of three trade-offs:

  • Business hours only, with a best-effort contact outside them. Honest, and fine if your customers do not need more.
  • Two or three people sharing a pager. This is what most companies have. It covers nights on paper, but one holiday or one resignation breaks it.
  • A smaller rotation with fewer people per shift, for example four engineers with a single person on call and a manager as backup. In our own costing, four engineers is the smallest in-house team that can run 24/7 at all.

Whichever you pick, write down what is covered and when, and say it the same way in your customer contracts.

Which on-call schedule works for a team of two to five?

Atlassian's on-call guide gives the simplest rules of thumb: a solo developer responsible for everything at all hours should bring in backup; teams of two usually alternate days; teams of three or more do well with weekly rotations. Every schedule should name a backup, because the primary can sleep through or miss a page.

Team size Schedule that works Watch out for
1 Business hours plus a named outside backup for emergencies No real cover: holidays and sick days leave nobody
2 Alternating days, or alternating weeks, each the other's backup Every second week on call; burnout within months
3 to 5 Weekly rotation, primary plus secondary, handover on a weekday morning Night pages still land on one person
Two or more time zones Follow-the-sun: each site covers its own daytime Handover quality and weekends

Team-size guidance from Atlassian; the right-hand column is our experience.

For a team of two, Atlassian describes one split where person A takes Monday and Wednesday, person B takes Tuesday and Thursday, and they alternate Friday to Sunday, so each gets every other weekend off. Alternating full weeks is the other common option.

What is the follow-the-sun model?

Follow-the-sun is an on-call schedule where each team is on call only during its own daylight hours and hands over to a team in another time zone at the end of its day. Atlassian defines it as "scheduling on-call rotations during daylight hours". Google's SRE book gives the reason it prefers multi-site teams once a service is big enough: night shifts harm people's health, and "a multi-site 'follow the sun' rotation allows teams to avoid night shifts altogether". The same chapter warns that multi-site teams "incur communication and coordination overhead" (source).

What it takes in practice:

  • People far enough apart. Two sites roughly 12 hours apart can each cover a 12-hour day. With three or more clusters, shifts get shorter and handovers more frequent.
  • A written handover. Open incidents, risky changes in flight and anything that may page in the next shift. A handover that lives in someone's head is the main way follow-the-sun fails.
  • The same access everywhere. Atlassian notes that distributed teams sometimes have access to different things, so you may still need an overnight backup, paged only for issues the other time zone cannot fix and that cannot wait until morning.
  • A plan for weekends. Follow-the-sun covers the working week. Weekends and public holidays usually run on a paged rotation.

A small team rarely has offices in two time zones. The usual ways to get follow-the-sun anyway are hiring one or two remote engineers far from your main time zone, or sharing a rotation with a support provider that already has one.

How do you set up an on-call schedule step by step?

  1. Decide what pages a person. Only problems that hurt users now and need a human. Everything else goes to a channel or a ticket. If your alerts come from an error tracker, tune those rules first; our DevOps best practices guide covers monitoring alongside the rest of the delivery pipeline.
  2. Set response times. Google gives typical values of 5 minutes for user-facing or time-critical services and 30 minutes for less time-sensitive systems. Pick yours per service and tell customers the same thing.
  3. Define primary and secondary. Google notes that teams split duties differently: some use the secondary as a fall-through for pages the primary misses, others have the primary handle only pages while the secondary handles non-urgent production work.
  4. Build the escalation policy in your paging tool. Primary, then secondary after a fixed number of minutes, then a manager. Test it with a real page.
  5. Pick a handover time. A weekday morning, when both people are awake and the team is around to answer questions.
  6. Write runbooks for the alerts that page. One page per alert: what it means, how to check it, how to mitigate it, who to call. Runbooks are much shorter when the environment is reproducible; see our guide to infrastructure as code.
  7. Agree on compensation. Google offers time off in lieu or cash, capped at a share of salary. The cap also limits how much on-call any one person takes on.
  8. Review monthly. Pages per shift, pages at night, and pages nobody needed to act on.

How much on-call load is too much?

Google sets a quantitative limit. It found that dealing with one on-call incident, including root-cause analysis, remediation and follow-up such as writing a postmortem and fixing bugs, takes about 6 hours on average. From that, "the maximum number of incidents per day is 2 per 12-hour on-call shift", and the distribution of pages should be flat with a likely median of zero (Being On-Call).

If your rotation regularly runs above that, adding people spreads the pain without reducing it. Look first at which alerts page, how many of them led to real work, and which system produces most of them. Teams on shared or hand-built infrastructure often find the fix is in the platform, which is the problem platform engineering sets out to solve.

How much does PagerDuty cost?

PagerDuty's official pricing page, read on 13 November 2026, lists two families: per-user Incident Management plans and Reliability Platform plans with a yearly platform fee.

PagerDuty plan Listed price What the page says it is for
Free $0, up to 5 users A single team trying on-call: 1 on-call schedule, 1 escalation policy
Professional $21 per user / month billed annually, $25 monthly Small teams standardizing incident management
Starter (PD Reliability Platform) $2,800 platform fee per year Small teams; adds AI noise reduction, status pages, post-incident reviews
Essential $12,000 platform fee per year Larger teams
Plus $48,000 per year Listed in the plan comparison
Ultimate Custom pricing Quoted on request

Source: PagerDuty pricing, accessed 13 November 2026. Prices change; check the page before you budget.

For a team of five, the Free plan covers the headcount but only one on-call schedule and one escalation policy. Professional on annual billing works out at 5 × $21 × 12 = $1,260 a year. Check which features you need, such as post-incident reviews or status pages, against the plan comparison on the page before you choose; some are add-ons purchased separately.

What does 24/7 on-call cost if you build it in-house?

The tool is the small part of the bill. People are the large one. This is our estimate for one mid-senior DevOps or SRE engineer, and for the smallest in-house team that can provide 24/7 cover.

Per engineer, per year United States United Kingdom
Base salary, market median $140,000 £80,000
Benefits, payroll taxes, pension $49,000 £18,450
On-call allowance, about 10% of base $14,000 £8,000
Tooling, training, equipment $5,000 £4,000
Fully loaded cost $208,000 £110,450
Minimum 24/7 team of four $832,000 £441,800

Valletta estimate for one mid-senior DevOps/SRE engineer. Assumptions are listed below the table.

Assumptions behind the figures:

  • US salary: $140,000 base, within the 2026 market composite for DevOps ($119,000 to $143,000) and SRE ($132,000 to $155,000) roles from Glassdoor, ZipRecruiter, PayScale and Built In, as compiled by KORE1 (2026).
  • US overheads: benefits and payroll taxes at 35% of base. The US Bureau of Labor Statistics reports benefits at 30% of total private-sector compensation (June 2026), equal to about 43% of wages, so 35% is conservative.
  • UK salary: £80,000, the median permanent SRE salary on IT Jobs Watch for the six months to 26 October 2026.
  • UK overheads: employer National Insurance at 15% above £5,000 (2026/27), 5% employer pension and £2,000 of other benefits.
  • On-call and tooling: on-call allowance at 10% of base; $5,000 or £4,000 per engineer for tooling, training and equipment.
  • Not included: recruitment costs, and the cost of replacing someone who leaves.

Four engineers is half of Google's single-site figure, so treat it as a floor rather than a target.

What does a healthy on-call rotation look like?

What good looks like

  • ✓Every shift has a named primary and a named backup, visible in the paging tool.
  • ✓Only alerts that need a human now can page; the rest go to a channel or a ticket.
  • ✓Response times are written down per service and match what customers were told.
  • ✓Pages per 12-hour shift stay at two or fewer, with most shifts at zero.
  • ✓Every paging alert has a runbook.
  • ✓Handover happens at a fixed time and is written down.
  • ✓On-call pay or time off is agreed and capped.
  • ✓The rotation survives one person on holiday without anyone working two weeks in a row.

If the last point fails, the rotation is too small for the promise you are making. Either narrow the promise, for example business hours plus best effort, or add people to the rotation from outside the team.

Need 24/7 cover without hiring four engineers?

Valletta Software Development runs DevOps and SRE support on a follow-the-sun rotation across five timezone clusters, so someone is mid-shift at every hour of the working week. A named engineer responds to a P1 within 15 minutes, with hourly updates until it is resolved. Targets are set per engagement and written into the contract, and support is priced from $45 per hour.

See DevOps support

Want the numbers for your own team first? Try the 24/7 cover cost calculator.

FAQ

What is an on-call rotation?

An on-call rotation is a schedule that makes one engineer, the primary, responsible for responding to production alerts during a set period, usually a day or a week, with a secondary as backup. Responsibility then passes to the next person. The rotation exists so urgent problems always have an owner without everyone being on call all the time.

How do you make an on-call schedule for a small team?

List who can respond, then pick a pattern by team size: alternating days for two people, weekly rotations for three or more, as Atlassian recommends. Name a backup for every shift, set response times, hand over on a weekday morning, and publish the schedule in your paging tool so escalation happens automatically.

What is the follow-the-sun model?

Follow-the-sun schedules on-call so that each team covers only its own daytime hours, handing over to a team in another time zone as its day ends. Google's SRE book notes that it lets teams avoid night shifts altogether, at the cost of coordination overhead between sites. It requires engineers in at least two distant time zones.

How many engineers do you need for 24/7 on-call?

Google's SRE book puts the minimum for a single-site team at eight, assuming a primary and secondary at all times, week-long shifts and at most 25% of time on call. For dual-site teams it suggests at least six per site. Teams smaller than that can still cover 24/7, but each person is on call more often.

How much does PagerDuty cost?

On 13 November 2026, PagerDuty lists a Free plan for up to 5 users, Professional at $21 per user per month billed annually or $25 monthly, and Reliability Platform plans from a $2,800 yearly platform fee, with Essential at $12,000 and Plus at $48,000. Ultimate is custom pricing, quoted on request.

How much should you pay engineers for on-call?

There is no single rule. Google's SRE book says it offers time off in lieu or cash, capped at a proportion of salary, and the cap also limits how much on-call one person takes. Our own cost model assumes an on-call allowance of about 10% of base salary. Whatever you choose, write it down.

Need DevOps hands for your infrastructure?

From containers to full CI/CD pipelines, our DevOps engineers keep production stable. Book a free 30 minute call to scope your setup.

Valletta.Software - Top-Rated Agency on 50Pros

Talk to the engineers behind this blog

Valletta Software builds and staffs dedicated development teams for companies across the EU and US. Tell us about your project and get a reply within one business day.