Is On-Call Even Necessary for Such a Small Team?
The immediate reaction when you hear “on-call” and “two people” in the same sentence is often a mix of pity and skepticism. Many assume it’s a setup for guaranteed burnout or, worse, for leaving customers stranded. But here’s the blunt truth: if your service is critical to your customers or your revenue, someone needs to be available when things go sideways. Even for a two-person team, an on-call rotation isn't just possible; it's a non-negotiable component of a professional operation. It’s about managing risk, setting expectations, and most importantly, preserving the sanity of those two people.
Think about it: Your customers don't care if you're a lean startup or a sprawling enterprise when their payment fails or your service goes offline. They just want it fixed. An unstructured 'whoever sees it first' approach quickly turns into a scramble, a blame game, or worse, a silent resignation to failure. Formalizing it, even minimally, creates a framework for responsibility and response.
What Does an On-Call Rotation for Two People Look Like?
At its core, a 2-person on-call rotation is a weekly (or bi-weekly) flip-flop. Person A is 'on' for a defined period, Person B is 'off' (or primary backup). Then they switch. Simple, right? Not quite. The simplicity of the schedule belies the complexity of making it sustainable.
- Primary/Secondary: One person is primary (responsible for initial response), the other is secondary (backup, for escalation or critical failure). This is crucial for truly 'off' time.
- Defined Handoffs: A clear point where responsibilities transfer. This isn't just about who holds the pager; it's about knowledge transfer – what's been happening, any ongoing issues, recent deployments.
- Shared Context: Both individuals must possess enough domain knowledge to handle most common incidents. If only one person understands a critical component, your 'rotation' is a fallacy.
- Respect for Off-Call Time: This is paramount. The person who is 'off' should genuinely be able to disconnect. No casual pings for non-emergencies.
A typical rotation might look like this:
- Week 1: Alice is Primary, Bob is Secondary.
- Week 2: Bob is Primary, Alice is Secondary.
If you're dealing with high-stakes incidents, a shorter rotation (e.g., 3-4 days) might seem appealing to prevent fatigue. However, for a two-person team, this means less time truly 'off' and more frequent context switching, which can itself be a source of stress.
Defining Your On-Call Scope: What's Truly Critical?
The biggest trap for small teams doing on-call is trying to be everything to everyone. You cannot. With two people, you must be ruthlessly selective about what warrants an alert and what can wait until business hours. This requires honest, sometimes uncomfortable, conversations about your service's real priorities.
Tier Your Incidents
- Critical (Pager-Worthy): Direct impact on revenue, user data, or core functionality. Your website is down, payments fail, user accounts are inaccessible. These warrant an immediate, out-of-hours response.
- High (Next Business Day): Significant user impact, but not a complete blocker. A specific report is delayed, a non-essential feature is buggy, a background job failed but can be re-run. These need attention, but not necessarily a 3 AM wake-up.
- Medium/Low (Scheduled Work): Minor bugs, UI glitches, performance degradation that doesn't halt operations. These go into the backlog.
Your monitoring tools need to reflect these tiers. A critical alert should trigger a pager (via PagerDuty, Opsgenie, or even a simple SMS if you're truly lean). A high alert might generate an email to a shared inbox, while lower priority items go directly to an issue tracker.
The Essential Toolkit for Lean On-Call
You don't need an enterprise-grade incident management system costing thousands per month. For two people, efficiency and clarity are key. Here are some tools that punch above their weight:
- Monitoring & Alerting:
- Sentry: Excellent for error tracking. Configure it to send critical errors (e.g., HTTP 500s above a certain threshold) to your pager. Costs start around $26/month.
- UptimeRobot/Statuscake: Simple external uptime monitoring. Essential for knowing if your site is genuinely down. Many free tiers are available for basic checks.
- Vercel/Cloudflare Logs: If you're using these platforms, their logging and analytics can be invaluable for diagnosing issues. Set up alerts for specific error patterns or usage spikes.
- Stripe/Payment Processor Status: Monitor their status pages. If Stripe is down, your payment failures aren't your fault, but you need to know.
- PostHog/Mixpanel: For product analytics. While not an alerting tool, seeing a sudden drop-off in key user actions (e.g., sign-ups, checkouts) can be an early indicator of a problem, even before technical alerts fire.
- Communication & Incident Management:
- Slack/Discord: A dedicated incident channel for communication during an active event. Keep it focused.
- Google Docs/Notion: A shared runbook. This is your bible. How to check logs, common fixes, escalation paths, contact details. Don't rely on memory at 3 AM.
- A Pager Solution: PagerDuty, Opsgenie, or even a simple Twilio SMS gateway for critical alerts. PagerDuty's free tier for small teams (up to 5 users) can be a lifesaver.
Remember, the goal is not to have more alerts, but to have smarter alerts. False positives are the fastest way to erode trust in your system and lead to alert fatigue.
Preventing Burnout: The Elephant in the Room
This is where most 2-person on-call rotations fall apart. Without careful management, it devolves into two people being perpetually 'on', leading to exhaustion, resentment, and ultimately, a failing service. Here's how to fight back:
1. Strict Boundaries
- Truly Off-Call: The person who is off-call should be unreachable for non-critical issues. If it's not causing immediate, severe impact on users or revenue, it waits.
- Handover Rituals: Don't just swap. Have a brief, mandatory handover meeting (even 15 minutes) to discuss any open issues, recent deploys, or potential risks. This builds confidence and reduces anxiety.
- Backup Plan for Backup: What if the on-call person is truly unavailable (sick, traveling without signal)? The 'secondary' needs to be ready to step up, but also acknowledge this is an exception, not the norm.
2. Post-Mortems (Even Small Ones)
Every significant incident, even if quickly resolved, warrants a quick debrief. What happened? Why? How can we prevent it? What can we automate? This isn't about blame; it's about learning and improving. This is how you reduce the load over time.
3. Automate, Automate, Automate
Your most valuable asset is time. Any manual check, diagnosis step, or recovery procedure that can be automated, should be. Can a script restart a service? Can a bot post status updates to your customers? Invest in these efforts during business hours.
“If you only have two people, every hour you spend on an incident is 50% of your operational capacity. Make every minute count.”
4. Scheduled Downtime & Maintenance Windows
If you need to deploy a major change or perform risky maintenance, schedule it during business hours. Announce it to users. This avoids a frantic 2 AM incident for something preventable.
When to Scale (and How SISL Can Help)
Eventually, if your business grows, a two-person on-call system will hit its limits. This usually happens when:
- Incidents become too frequent, impacting development time.
- The complexity of your system outstrips two people's ability to maintain full context.
- Burnout becomes a persistent issue, affecting team morale and performance.
At SISL, we often advise growing startups on establishing robust, yet lean, operational processes. We understand the tight constraints of small teams and can help you identify your critical paths, set up appropriate monitoring, and define incident response protocols that scale without immediately requiring a 24/7 internal ops team. Sometimes, the right move isn't hiring more, but optimizing what you have or smartly outsourcing specific monitoring needs.
If you're grappling with the complexities of keeping your service reliable without overwhelming your small team, perhaps it's time for a fresh perspective. We're here to help you build resilient systems that support your growth, not hinder it. Feel free to get in touch to discuss your unique challenges.
Final Thoughts
Running an on-call rotation with just two people is a balancing act of discipline, smart tooling, and unwavering respect for each other's time. It requires a clear understanding of what truly constitutes an emergency and a commitment to continuous improvement. Done right, it ensures your customers are supported, your service remains reliable, and your small team avoids the rapid descent into chronic exhaustion. It's not about being heroes; it's about being pragmatic and sustainable.