Lorixen.
EN

A fair on-call rota for a small tech team

Short answer: If your business runs a website, app or online store that customers use around the clock, someone needs to be reachable when it breaks at 2 a.m. For a small team, a fair on-call rota means clear hours of cover, a simple weekly rotation with a backup, alerts that only page for real customer-facing problems, a short guide for common failures, a clean handover, and time off or pay for nights that went badly. The goal is a rota people can sustain for years, not a hero who answers every alert.

Do you need out-of-hours cover at all?

Start with the business question. What happens if your site, booking system or app is down from Friday evening until Monday morning? For a brochure site, maybe nothing serious. For an online shop during a sale, a delivery app or a SaaS product with customers in other time zones, it could mean lost orders and broken trust.

Options range from no out-of-hours cover, with a clear status message and monitoring that alerts the team in the morning, to evening and weekend cover for peak periods, to full 24/7 cover. Many small teams sensibly choose cover only for business-critical systems, such as checkout and payments, and leave everything else for working hours.

Atlassian's guide to on-call describes the same choice: team members take turns staffing a rotation, either around the clock or only outside normal business hours.

How big does the team need to be?

Google's Site Reliability Engineering book, in its chapter on being on-call, gives a useful reference point. It describes a rule that no more than 25% of an engineer's time should be spent on call, and, assuming two people on call at once (a primary and a secondary) with week-long shifts, derives a minimum of eight engineers for 24/7 cover from a single-site team.

Most small businesses do not have eight engineers, and that is the point: with three or four developers, full 24/7 cover will be heavy. Ways to make it workable:

  • Limit cover to the systems that genuinely need it.
  • Cover evenings and weekends only during peak trading periods.
  • Share cover with an agency or managed hosting provider for infrastructure issues.
  • Reduce the number of things that can wake people up, which is often the biggest lever.

Set up the rotation

A simple, predictable pattern beats a clever one:

  • Weekly shifts, handing over at the same time each week, ideally mid-morning on a weekday rather than Friday evening.
  • A primary and a backup. The SRE book notes that teams use the secondary differently, for example as a fall-through for pages the primary misses. For a small team, the backup is mainly there if the primary does not acknowledge an alert within a set time.
  • An escalation path, such as primary, then backup, then the technical lead, configured in your alerting tool so it happens automatically.
  • Swaps made easy, with a shared calendar and a simple rule for trading weeks.
  • No back-to-back weeks unless someone asks for it.

Only page for what matters

An on-call week is only fair if the pager is quiet most of the time. The SRE book argues that the distribution of paging events should be flat with a likely median of zero, and estimates that handling an incident, including follow-up such as a postmortem and fixes, takes about six hours, putting a ceiling of about two incidents per 12-hour shift.

Practical rules:

  • Page only for problems that affect customers now, such as checkout failing, the site down or payments erroring.
  • Send everything else, such as disk filling slowly or a background job running late, to a channel or ticket queue for working hours.
  • Review every page after the week: was it real, was it urgent, could it have waited?
  • Fix the causes of repeated pages rather than living with them.

Equip the person on call

  • A short guide for common problems, such as how to restart the app, roll back a deploy or switch payments to a backup provider.
  • Access that works from home, tested before the shift: VPN, admin logins, hosting dashboard, MFA device.
  • Contact details for the hosting provider, payment provider and the business owner.
  • Permission to act, including rolling back a release or putting the site into maintenance mode without waiting for approval.

Handover

At the end of each shift, the outgoing person writes a few lines: incidents, anything still unresolved, alerts that should be tuned, upcoming risky events such as a sale or a planned migration. The incoming person reads it and asks questions. Five minutes of handover prevents a lot of confusion.

Fair reward and recovery

The SRE book notes that adequate compensation needs to be considered for out-of-hours support and that organisations handle it differently, for example time off in lieu or cash. Agree a simple, written approach for your team: a stipend per week on call, extra time off after a night incident, or both. Make the rule that someone woken at night starts late the next day, no questions asked.

A worked example

Picture a hypothetical Denver online garden-supplies retailer with three developers and a busy spring season. One developer had informally answered every alert for two years and was burning out.

The team agreed cover only for checkout, payments and the order system, with evening and weekend cover from March to June and best-effort the rest of the year. They set up a weekly rotation with a backup, automatic escalation after 15 minutes, and a written guide for the five most common failures. Non-urgent alerts moved to a daytime channel. After each week they reviewed pages and removed noisy ones. The number of night-time pages fell, the original developer took a full turn off, and the next season went without anyone dreading their week.

FAQ

Can the business owner be on the rota?

Yes, as an escalation contact for business decisions, even if not for technical fixes.

Do we need paid alerting software?

Not at first. Many monitoring tools can call or text. Paid tools add escalation, schedules and acknowledgement tracking.

How long should shifts be?

A week is common and easy to plan. Shorter shifts suit teams where night pages are frequent.

If one person on your team ends up answering every alert at night, tell us what you monitor today.

Sources

WhatsAppBook a call