Engineering
Write the incident review from what happened
Alerts, the incident channel, deploys and tickets in one timeline, with every line linked, so the review starts from facts.
What happened during Tuesday's checkout outage?
- 1PagerDuty · 23 alerts✓ read
- 2Slack · #inc-482 316 messages✓ read
- 3GitHub · 4 deploys✓ read
Checkout failed for 41 minutes after a config change raised the payment timeout to 60 seconds. The first customer ticket came 6 minutes before the first alert.
The result
3 h → 25 min
to a first review draft
Time from incident close to a first review draft at a 30-engineer team, median of five incidents (demo figure).
Timelines are archaeology
Reconstructing an incident means scrolling a channel, matching timestamps to deploys and finding the first report. Skein builds the timeline from the sources and links each line, so the people in the review argue about causes, not times.
How teams use it
A first draft of every review, with the timeline filled in.
What did customers see, and when did we first hear about it?
Has this happened before?