What execs should demand before the drill

This is not admin. This is control.

  • One timeline (minute-by-minute) with owners, not swimlanes and hope
  • One decision owner for go/no-go and rollback
  • One comms channel for the night (plus a backup)
  • One source of truth folder: runbooks, scripts, checklists, screenshots
  • One list of “do not touch” systems during freeze (and who enforces it)

The cutover storm drill checklist

1) Define “pressure” before you start

If you cannot describe failure, you cannot practise recovery.

  • What counts as a bad outcome: missed trades, bad positions, failed settlements, broken interfaces, corrupted master data
  • What you protect first: trading continuity, cash, regulatory reporting, data integrity
  • What you will accept: a degraded mode, a reduced product set, manual workarounds (with limits)

Output: a one-page “failure modes” list with owners.

2) Set the relief valve: rollback triggers that actually trigger

Most rollbacks fail because they are vague or political.

Write triggers as facts:

  • “Interface X fails for Y minutes”
  • “EOD batch does not complete by time T”
  • “Position reconciliation variance exceeds threshold”
  • “Confirmation generation error rate exceeds N%”
  • “ERP posting backlog exceeds N items”

For each trigger, specify:

  • who calls it
  • what evidence they use
  • how long they wait
  • what the rollback action is

Output: a signed rollback table.

3) Rehearse the cutover steps end-to-end

Do not let the rehearsal stop at “migration done.”

Run the full chain:

  • freeze start and access controls
  • data load / configuration promotion
  • interfaces on/off sequencing
  • smoke tests on day-one workflows
  • monitoring checks and alert routing
  • handover to support with named on-call coverage

Output: a rehearsal log with timestamps and pass/fail.

4) Rehearse comms like it’s the real night

Silence causes panic. Panic causes bad changes.

Set a simple comms rhythm:

  • every 30 minutes: current step, next step, risks, decisions needed
  • one message template for issues: “what broke, impact, owner, ETA, workaround”
  • one escalation path: who gets pulled in, when, and why

Output: comms transcript plus a cleaned “playbook” for go-live night.

5) Prove operability: the handover that keeps January clean

A successful cutover still fails if Ops inherits confusion.

Confirm before you close the rehearsal:

  • runbooks exist for the top 5 incidents
  • monitoring exists for the top 10 integrations
  • roles and permissions match real job functions
  • support model is active (L1/L2/L3, named people, roster)

Output: a signed support readiness checklist.

Outcomes, risks, mitigations

Short-term outcomes

Faster decision-making during go-live weekend

Fewer surprise dependencies

Clear rollback confidence, which reduces stress

Risk: people treat rehearsal as a demo.

Mitigation: pass/fail logs only. No “nearly.”

Long-term outcomes

Less post-go-live fatigue because incidents drop and fixes become planned

Fewer finance and reconciliation shocks in the first close after go-live

Risk: teams skip the ugly scenarios.

Mitigation: force one “break” scenario per rehearsal (interface down, bad data, role mismatch).

Unexpected outcomes

Vendor behaviour changes when rollback triggers are explicit

Internal ownership improves when the decision rights are written, not assumed

Risk: politics around who owns the call.

Mitigation: name the decision owner before the drill. Do not negotiate at 02:00.

Five-step rollout plan

  • Lock day-one workflows and failure modes.
  • Write rollback triggers as measurable facts.
  • Run the rehearsal with a minute-by-minute timeline.
  • Capture evidence: logs, screenshots, timestamps, decisions.
  • Publish the “go-live night pack” and rehearse comms.

Three questions to tune this to your program

  • What is your freeze window and what systems must stay stable (CTRM, ERP, market data, scheduling, EDI, banking)?
  • What is the single business failure you refuse to tolerate on day one: wrong position, failed settlement, wrong invoice, or delayed confirmations?
  • Who holds final authority for go/no-go and rollback: COO, CIO, Head of Trading, or a joint gate?
Published On: January 10th, 2026 / Categories: Advisory, Transformation /