Part 3 · 1 chapters · ~12 min

Incident Response

Running an incident with structure: declaring early, the commander who coordinates, operations making announced changes, mitigation before diagnosis, communication on a cadence, explicit resolution and handover, with severity definitions and customer wording for a fintech.

6

Roles, severity, communication and runbooks

run it, do not fight it
  1. Declare early, with a severity, a channel and a timeline document.
  2. The commander coordinates and does not debug.
  3. Operations: one change at a time, announced before it is made.
  4. Mitigate first: roll back, fail over, switch, disable, shed load.
  5. Communications on a fixed cadence. Silence is the thing users do not forgive.
  6. Resolve explicitly, then assign the postmortem and file follow-ups.
severitydefinition (a fintech)responsecomms
SEV1money movement down or wrong for many users; data exposure; regulatory impactall hands, exec informed, 24/7 until resolvedstatus page within 15 min; updates every 30 min; regulator notified per rules
SEV2a key journey degraded (transfers slow, one bank failing); budget burning faston-call + owning team; commander assignedstatus page if customer-visible; updates hourly
SEV3minor degradation, workaround exists, no money at riskowning team in working hoursinternal channel
SEV4cosmetic or internal-onlyticketnone
code
STATUS PAGE (customer-facing, plain words)
Investigating: Transfers to some banks are delayed
Some transfers to GTBank and Access Bank are taking longer than usual to arrive.
Your money is safe. Transfers that are delayed will either complete or be reversed
automatically, so please don't retry them. Next update by 14:30 WAT.
the frontend's role in an incident
The app is the first place users see an incident. The Trust course part 7 designs the degraded banner and the "confirming" states, and the incident runbook includes the flag that turns the banner on. Turning on the banner is often the very first mitigation.
RUNNING AN INCIDENT
from the page to resolution: declaring, roles, mitigation first, communication, and handing back
swipe the figure sideways, or tap expand for full screen
1/6
declare
Declare early: the on-call engineer is paged, confirms user impact, and declares an incident with a severity, opening a channel (#inc-2026-10-06-transfers) and a timeline doc. Declaring is cheap and can be downgraded; under-declaring delays help. If in doubt, declare.