On-Call Fairness Analyzer
Analyze your on-call rotation for fairness and burnout risk. See pages per engineer, weekend distribution, and a weighted burnout score that flags overloaded teammates.
For your own reference — not used in calculations.
| Engineer | Total pages | After-hours | Weekend | |
|---|---|---|---|---|
Top engineer was paged 3.1× more than the lightest. Investigate whether a specific shift, customer segment, or alert pipeline is overweighted.
Total pages (Last 30 days)
89
17.8 avg per engineer
After-hours
34
38% of pages
Weekend / holiday
20
22% of pages
Per-engineer burnout score(total + 2× after-hours + 3× weekend)
Automate your incident response
Reduce MTTR by 90% with AI-powered root cause analysis. Free to start.
On-call burnout is the SRE attrition machine
Engineers do not usually quit because of one bad week. They quit because the load was uneven, the after-hours pages stacked up, and the weekend shifts kept landing on them while the rest of the rotation looked lighter from the outside. The first job of any on-call review is to surface that pattern in numbers — and the second is to fix it.
What good rotation health looks like
- Fairness ratio under 1.5× over a 90-day window. Some week-to-week variance is unavoidable, but quarter averages should converge.
- After-hours pages under 30% of total. If most pages are after-hours, you are alerting too aggressively on issues that could wait for business hours.
- Weekend pages under 15%. If weekends are heavier than weekdays, traffic patterns or alert thresholds are mistuned.
- No engineer above 2× team median — a single overloaded person is your earliest churn signal.
Three structural fixes
1. Cut alert noise first. The cheapest reduction in on-call load is removing pages that did not need to be a page. Use the Alert Noise Score tool to identify candidates.
2. Rotate "primary" and "secondary" roles. Have a primary always-on engineer plus a secondary who handles overflow, then rotate both. Halves the cognitive load of being on-call without doubling headcount.
3. Automate the first 30 minutes. Most incidents follow predictable patterns — gather context, check recent deploys, run health checks. This is exactly what Uptimes.ai automates: when an alert fires, the AI SRE agent investigates and only escalates if it cannot resolve. For many incident classes, the human is never paged.