A cost-cutting review flagged our weekly on-call handoff meeting as low-value overhead. We nearly cancelled it before checking what it actually did.

Fifteen minutes, once a week, outgoing on-call engineer talks to incoming on-call engineer, camera optional. On an efficiency audit of recurring meetings, it looked like an easy fifteen minutes to reclaim. We almost cut it before deciding to actually measure what it was worth first.

What we measured before cutting it

We ran a four-week experiment: two rotation groups kept the live handoff, two switched to async-only written notes for the same information. We tracked time-to-context for the incoming engineer on their first page of the week, and a post-shift survey question about confidence going into the rotation.

experiment: live_handoff_vs_async_only
groups: A (live, 15min) vs B (async notes only)
metric_1: time_to_context_first_page_min
  A: median 3.5
  B: median 9.0
metric_2: pre_shift_confidence (1-5 self-rated)
  A: mean 4.1
  B: mean 3.2
metric_3: meeting_time_cost_per_week_min
  A: 15  |  B: 0

What the written notes couldn't capture

The async notes had all the same facts. What they didn't have was tone — whether the outgoing engineer said "this alert is annoying but harmless" versus "this alert scares me and I never got to the bottom of it," a distinction that changes how urgently the incoming engineer treats the same line item, and one that rarely survives being written down formally.

"You can write 'still investigating elevated latency' in a doc. You can't write the exact worried tone in someone's voice that tells you to actually watch it closely tonight."

Why we kept it, and what we trimmed instead

We kept the fifteen-minute live handoff and found the actual savings elsewhere — cutting a redundant Friday status meeting that overlapped almost entirely with information already in the incident channel. The lesson wasn't "meetings are good" or "meetings are bad." It was that a meeting's value has to be measured against what it specifically enables, not judged by its length on a calendar.

  • Before cutting a recurring meeting for efficiency, measure what it specifically produces — tone and confidence rarely survive being converted to text.
  • Async handoffs are cheaper but measurably slower to get someone oriented; the tradeoff isn't free even when it looks free.
  • Look for savings in genuinely redundant meetings instead of the ones that quietly carry information nothing else replaces.