Page Arrival Time
P50 pager arrival latency.
Overview
Page arrival time is the latency between alert firing and engineer notification. Without measurement, slow page delivery hides as “the engineer was slow to ack”; with measurement, carrier delays and channel issues become operational signals.
- p50 pager arrival latency. Median latency per alert from fire to notification. The headline number; tells you whether the pipeline works at all.
- Per-channel latency. SMS, email, push, phone call. Each channel has different latency characteristics.
- Per-region carrier delay. Different carriers add different delays. Global teams especially see this.
- Per-severity SLO plus quarterly audit. Sev-1 needs sub-30-second arrival; Sev-3 can tolerate minutes; quarterly audit catches drift.
The approach
Three habits keep page arrival predictable: monitor per channel, SLO per severity, and audit quarterly.
- Per-channel monitoring. Track arrival latency per delivery channel. The dashboard surfaces which channel is the laggard.
- Per-severity SLO. Different severities tolerate different latencies. Document the target per tier so the team knows what to defend.
- Per-region carrier review. Quarterly check on carrier delays per region. Some carriers degrade silently over time.
- Quarterly audit plus documented target. Walk arrival latency each quarter; per-team SLO lives in the runbook.
Why this compounds
Each measured arrival supports investigation when the next page arrives slowly. Compounded across the year, the team has data instead of guesses about why MTTA drifted.
- Faster incident response. Right arrival produces fast response. MTTA stops being a black box.
- Operational fit. Right channel matches priority. Sev-1 paths optimised for speed; lower severities can tolerate slower channels.
- Cultural signal. Arrival monitoring signals that on-call matters. Engineers see the team protecting the rotation.
- Year-one investment, year-two habit. The first measurement is investment. By year two the metric is part of the standard review.