Follow the Sun On-Call: How to Set It Up
Set up follow the sun on-call across regions: staffing math, UTC handoff times, a daily handoff ritual and escalation that stays in the awake region.

Follow the sun on-call is the rotation model where on-call duty moves between engineers in different regions so that whoever holds the pager is working during their own daytime. Done well, nobody gets woken at 3am for a routine page, incidents get handled by people who are awake and sharp, and the night shift simply stops existing. Done badly, it becomes three half-owned rotations stitched together with a shared doc nobody reads, and incidents fall through the cracks at every handoff.
This guide covers how to set up follow the sun on-call in practice: when you actually have enough people to do it, how to choose handoff times, how to design the handoff itself, how escalation should work across regions, and how to run it inside Slack so the rotation stays visible. If you want to sketch a schedule while you read, the on-call schedule generator will build one and export it to your calendar.
What Follow the Sun On-Call Actually Means
In a traditional single-region rotation, one engineer holds the pager for a full week, 24 hours a day. They handle daytime pages at their desk and night pages from bed. In a follow the sun model, each day is split into windows, and each window belongs to a region where that window falls inside normal working hours.
A typical two-region setup looks like this:
- Region A (for example Europe): holds the pager from 07:00 to 19:00 UTC.
- Region B (for example US West): holds the pager from 19:00 to 07:00 UTC, which is roughly 11:00 to 23:00 Pacific.
A three-region setup (for example India, Europe and US) can use three eight-hour windows, which lets every shift sit comfortably inside the working day.
Why Teams Bother
The case for it is mostly about people. Google's Site Reliability Engineering book, in its chapter on being on-call, notes that night shifts have detrimental effects on people's health and that a multi-site follow the sun rotation lets teams avoid night shifts altogether. That is the core promise. Engineers who sleep through the night make better decisions during the day, stay in the rotation longer, and are less likely to quietly start looking for another job.
There is also a quality argument. An engineer paged at 03:00 local time is working with a fraction of their usual judgement. They are more likely to apply a risky fix, skip a step in the runbook, or forget to tell anyone what they did. A follow the sun model puts every page in front of someone who is at their desk, caffeinated, and surrounded by colleagues they can pull in.
What It Costs
The same SRE chapter is clear that multi-site teams incur communication and coordination overhead. Every handoff is a point where context can be lost. Every region needs enough engineers to make its own window sustainable. And the regions need to agree on tooling, severity definitions and escalation rules, or the rotation fragments into several local cultures that do things differently.
Follow the sun is not free. It trades night pages for handoff discipline. If your team is not willing to invest in the handoff, you are better off with a well-run single-region rotation.
Do You Have Enough People for Follow the Sun?
This is the first question to answer, and the most common reason a follow the sun setup fails. Splitting a five-person team across two continents does not give you follow the sun. It gives you two tiny rotations that each burn out faster than the original one.
The Staffing Math
Google's SRE guidance gives a useful anchor. It suggests that no more than 25 percent of an SRE's time should be spent on-call, and from that it derives a minimum of eight engineers for a single-site 24/7 rotation with a primary and a secondary. For dual-site teams, it suggests a reasonable minimum of six engineers per site.
You do not need to adopt Google's exact numbers to use the logic. The questions to ask for each region are:
- How many engineers in this region are genuinely able to respond to production incidents for the services in scope?
- If each of them takes one window shift per rotation cycle, how often does their turn come around?
- Is there a secondary or backup in the same region for every window, or does the backup live in a region that is asleep?
- What happens to the rotation when one person is on holiday and another is sick in the same week?
As a rough rule, if a region has fewer than four people who can credibly hold the pager, that region is not ready to own a window on its own. It can still participate as a secondary or as a shadow while it grows.
When a Small Team Should Not Try
If you have a strong engineering hub in one region and two or three remote engineers elsewhere, resist the temptation to hand the remote engineers the entire night. They will become the de facto night shift for the whole company, without the support, context or authority the main team has. That is not follow the sun. That is outsourcing the pain.
For teams in that position, a better first step is usually to reduce night pages at the source (stricter alert thresholds, fewer non-actionable alerts) and run a single-region rotation with a fair compensation policy. Our guide to 24/7 on-call coverage with a small team walks through the patterns that work at low headcount.
How to Set Up Follow the Sun On-Call, Step by Step
Once you know you have the people, the setup comes down to five decisions: which services are in scope, where the window boundaries sit, how the handoff works, how escalation crosses regions, and how the whole thing is made visible.
Step 1: Define the Scope
Start by listing the services the rotation covers. Follow the sun works best when every engineer in the rotation can meaningfully respond to every service on the list. If one region only knows the payments stack and another only knows the data platform, you do not have one follow the sun rotation. You have two specialised rotations with a gap.
Write down, for each service:
- The runbook or first-response doc, and whether it is current.
- The dashboards a responder should open first.
- Which regions have engineers who have handled a real incident on it in the last six months.
- The subject matter expert to escalate to, per region if possible.
Services that only one region understands should either get cross-training before launch or stay on a separate specialist rotation with its own escalation path.
Step 2: Choose Handoff Times
Handoff times matter more than people expect. The goal is for each handoff to happen when both the outgoing and incoming engineers are awake and at work, ideally with some overlap.
Practical guidelines:
- Anchor in UTC. Define every window in UTC in your scheduling tool, then let each person's calendar show it locally. This removes most of the confusion around daylight saving.
- Plan for daylight saving drift. Europe and North America change clocks on different dates, and India does not change at all. A handoff that sits at 09:00 in one city for most of the year can slide to 08:00 or 10:00 for a few weeks. Pick boundaries that remain inside working hours on both sides even after the drift.
- Avoid the very start and end of the day. A handoff at 08:00 local catches people on the commute. A handoff at 18:30 catches them leaving. Mid-morning for one side and late afternoon for the other is usually the sweet spot.
- Build in overlap. Even 30 minutes of overlap, where both engineers are reachable, makes a handoff during an active incident far safer.
Step 3: Design the Handoff
The handoff is where follow the sun lives or dies. A handoff is not the moment the pager changes hands. It is the moment context changes hands, and that needs a deliberate ritual.
A good handoff note answers four questions:
- What is open? Active incidents, their severity, who is working them, and the next action.
- What is noisy? Alerts that fired repeatedly and were judged non-actionable, so the next person does not chase the same ghost.
- What changed? Deploys, config changes, feature flags or infrastructure work in the last window that could plausibly cause trouble.
- What is coming? Planned maintenance, launches, vendor work or known risky jobs scheduled in the next window.
Keep the handoff in the same place every day. A dedicated Slack channel or a pinned thread in the on-call channel works well, because it is searchable, timestamped and visible to everyone. A private DM between two engineers is the worst option: the moment one of them is unavailable, the context is gone.
Step 4: Decide How Active Incidents Cross a Handoff
The hardest case is an incident that is still burning when the window ends. Teams typically choose one of two policies:
- Owner stays until stable. The outgoing engineer keeps the incident until it reaches a defined stable point (mitigated, customer impact stopped), then hands it over with a written summary. The incoming engineer takes all new pages immediately.
- Clean transfer at the boundary. The incident moves with the pager, and the outgoing engineer stays on a live call for a short, agreed period to brief the incoming one.
Either works. What does not work is leaving it undefined, because then every long incident becomes a negotiation at the worst possible moment. Write the policy down, and give it a time limit so the outgoing engineer is not quietly working a 14-hour day. For major incidents, it often makes sense to hand over the incident commander role explicitly as well, stating it in the incident channel so responders know who is in charge.
Step 5: Set Up Cross-Region Escalation
Escalation in a follow the sun rotation needs one extra rule compared with a single-region rotation: the escalation path for each window should, wherever possible, stay inside the region that is awake.
A sensible structure looks like this:
- Primary: the on-call engineer for the current window.
- Secondary: another engineer in the same region, also in working hours.
- Tertiary: an engineering lead or subject matter expert, still in the same region if possible.
- Cross-region: only for genuine high-severity incidents, wake an engineer in another region who owns the affected system.
The last step is important. Follow the sun does not mean nobody is ever woken up. It means waking someone up becomes a deliberate decision for severe incidents rather than the default for every page. Define which severities justify a cross-region page, and make sure the people who might be woken know that in advance.
Running Follow the Sun On-Call in Slack
Most of the coordination cost of follow the sun comes from people not knowing who is on call right now. When the answer changes two or three times a day, a static wiki page goes stale within hours. The rotation needs to be visible where the work already happens.
Make the Current Responder Obvious
Use a single handle, such as an @oncall Slack user group, that always points at whoever holds the pager in the current window. Anyone in the company can then mention @oncall without checking a timetable or thinking about time zones. When the window changes, the group membership changes with it. Pagerly's rotations in Slack handle this pattern: you create the rotation once, and Pagerly keeps the user group in step with the schedule.
Post the Handoff Automatically
Make the handoff announcement automatic: a message in the team channel at each window boundary saying who is going off call and who is coming on. That message becomes the natural place to drop the handoff note as a thread reply. Over time, the channel turns into a complete, searchable log of what each region saw and passed along.
Page the Right Region
Paging should follow the same windows as the Slack handle. When an alert fires, it should reach the engineer on shift through the channels they actually notice: Slack, mobile push, SMS or a phone call for the most severe cases. Pagerly's on-call paging covers email, Slack, SMS, phone call and app notifications, so you can keep low-severity alerts in Slack and reserve phone calls for incidents that justify waking someone.
Common Follow the Sun Mistakes
Treating Each Region as Its Own Team
When each region runs its own retrospectives, writes its own runbooks and defines severity its own way, the rotation stops being one system. Incidents look different depending on which half of the day they start in. Keep one set of severity definitions, one runbook library and one retrospective process, and invite people from every region to every review of a significant incident.
Letting One Region Absorb the Weekend
Weekends are where fairness tends to slip. If one region always covers Saturday because "it is quieter there", that region carries a hidden extra load. Rotate weekend responsibility across regions, or split weekend windows the same way as weekdays, and track who covered what.
Skipping the Handoff When Things Are Quiet
Quiet days are exactly when teams stop writing handoff notes, and then the first busy day exposes that nobody remembers how. Make the handoff mandatory even when the note says "nothing open, nothing noisy, no changes". That one line still confirms the incoming engineer is present and has taken the pager.
Ignoring Holidays That Do Not Overlap
Public holidays differ between countries. Diwali, Thanksgiving, Golden Week and Christmas fall on different days and affect different regions. Put every region's public holidays into the scheduling tool early, and plan cover swaps weeks ahead instead of discovering on the day that a whole region is offline.
Measuring Only Response Time
Follow the sun usually improves acknowledgement time, because nobody is fumbling for a phone in the dark. But the more important metrics for this model are how many pages were cross-region escalations, how many incidents spanned a handoff, and whether those incidents took longer to resolve. If incidents that cross a handoff consistently take longer, the handoff is the thing to fix.
A Practical Rollout Plan
Switching a live rotation to follow the sun in one step is risky. A staged rollout gives you time to find the gaps.
Weeks 1 to 2: Shadow
Keep the existing rotation as the official one. Add engineers from the new region as shadows for their future window. They receive pages, join incidents and practise the handoff, but the original on-call engineer still owns the response. Collect every question the shadows ask. Each one is a missing runbook entry or a missing piece of context.
Weeks 3 to 4: Reverse Shadow
Swap roles. The new region holds the pager for its window, with the original engineer as an explicit backup. Escalations to the backup are expected and fine; the point is to learn which situations still need them.
Week 5 Onward: Go Live and Review
Make the follow the sun rotation the official one, with in-region secondaries. Hold a short review after the first month covering handoff quality, cross-region escalations, and how each region feels about its load. Adjust window boundaries if one side is consistently handling more pages than the other.
Is Follow the Sun Right for Your Team?
Follow the sun on-call is one of the most effective ways to make on-call humane, but only when its preconditions are in place. Before committing, check the following:
- You have enough engineers in each region to staff a window, plus a secondary, without anyone taking a turn every few days.
- Each region can respond to every service in scope, or you have a clear specialist escalation for the ones it cannot.
- Handoff times sit inside working hours on both sides all year, including daylight saving changes.
- You have a written handoff format and a single place where handoffs live.
- There is a defined policy for incidents that cross a window boundary.
- Cross-region escalation is reserved for genuinely severe incidents, and everyone knows which ones.
- The current on-call responder is visible in Slack through a single handle, updated automatically.
If most of those are true, follow the sun will likely give your team calmer nights and faster, better-judged responses. If several are missing, fix those first. A well-run single-region rotation beats a poorly run global one.
Conclusion
The appeal of follow the sun on-call is simple: let the planet's rotation do the work, so every page lands on someone who is awake. The reality is that the model moves effort from night pages into handoffs, staffing and shared standards. Teams that invest in those three things get the benefit. Teams that skip them get the coordination overhead without the rest.
Start with honest staffing math, anchor your windows in UTC, make the handoff a daily ritual that lives in a shared channel, keep escalation inside the awake region by default, and make the current responder visible to everyone through a single Slack handle. Roll it out in stages, measure the incidents that cross a handoff, and tune from there. When you are ready to put it into practice, sketch your windows with the on-call schedule generator, then run the rotation where your team already works.
