Why Every Go-Live Should Be Evidence Driven, Not Slide Driven - Banking Merger Scenario
- Sunil Dutt Jha
- Nov 10
- 4 min read
Banking Merger Scenario - How a perfect readiness plan failed in the first hour and what evidence-driven GO-LIVE change. The steady way to stop “It passed testing, but died at cutover” failures — by wiring evidence to ownership, not slides
Two major retail banks merged to form a single digital-first entity. Every readiness meeting said “green.” Testing cycles passed. The integration dashboards showed 99.7% completion. But within hours of go-live, the first-hour ledger failed to reconcile across merged customer accounts. Balances vanished temporarily, credit limits duplicated, and branch operations froze for two days.
Nothing was wrong with technology — the drift happened between P3 (logic rules) and P5 (implementation ownership). What looked like a system issue was actually an anatomical misalignment: evidence of readiness wasn’t owned, it was only presented.
Executive Context — Why This Keeps Happening
Every transformation program celebrates its “all-green” dashboard. Tests passed, sign-offs done, confidence high. Yet, hours after go-live, reports stop reconciling and exceptions flood in. The reason isn’t incompetence — it’s anatomy.
The enterprise passed testing on function, but not on ownership. Evidence lived in slide decks, not in verifiable artifacts.
To end the cycle, readiness must be evidence-driven, not meeting-driven. And that means wiring every test, rule, and map back to a living owner.
Quick Refresher: ICMG Enterprise Anatomy™
P1 – Strategy: Direction and value outcomes to achieve.
P2 – Process: Sequence of business activities to realize the strategy.
P3 – Systems / Logic: Data/rule/function/network/channel systems that execute the flow.
P4 – Component Specifications for IT: APIs, datasets, tables, screens, configurations.
P5 – Implementation Tasks: Human/IT tasks that build, modify, deploy components.
P6 – Operations: Business/IT ops that run, monitor, and maintain the service.
The Anatomy of the Stall
When projects collapse at cutover, the failure sequence looks identical:
P1 says “synergy/benefit.”
P2 outlines target journeys.
P3 maps (customer/product/limit/GL) are partial.
P4 runbooks define windows, not rules.
P5 tracks ceremonies, not verifiable evidence.
P6 absorbs defects that were predictable in P3.
Result: The First-Hour Ledger doesn’t reconcile with promises made in P1–P2.
Make Go/No-Go Evidence-Driven
The remedy isn’t more rehearsal; it’s structural proof.
Run a 48-hour pre-cutover P3 simulation that produces a First-Hour Ledger — a deterministic, queryable artifact your P5 owners must sign.
Go/No-Go becomes data, not opinion.
What the 48-Hour P3 Simulation Does
In the banking merger case, the fix came from replaying 30 days of pre-merger transactions through the new unified core.
That 48-hour simulation produced the First-Hour Ledger — exposing the exact 0.3% of accounts where credit-limit rules overlapped or settlement logic drifted.
By binding each diff line to an owner, the next release went live with zero GL variance.
Inputs
Customer map (who is in/out, segments, KYC/KYB flags)
Product map (SKUs/plans, states, version gates, entitlements)
Limits map (credit/transaction/throughput, per-channel/per-jurisdiction)
GL map (accounting events, cost centers, tax codes)
Actions
Re-run the last 30 days’ representative traffic through the new rule sets.
Generate the First-Hour Ledger: all expected postings, rejects, exceptions, throttles.
Emit a Diff Pack vs current production (rows, amounts, counts, reasons, owners).
Bind to P5 Owners
Each diff line must have an owner in P5 (name, team, SLA) and a closure state (fixed / accepted / guardrailed).
Go/No-Go = First-Hour Ledger status, not meeting minutes.
If it’s red, there is no cutover.
Minimal Artifact Set — Use This Exact Bundle
A1 – P3 Maps (versioned): customer.yaml, product.yaml, limits.yaml, gl.yaml
A2 – Simulation Notebook: inputs → rules → outputs (with seed & timestamp)
A3 – First-Hour Ledger: {event_id, posting, amount, GL, rule_version, outcome}
A4 – Diff Pack: {old_vs_new counts, $ deltas, exception classes, top 10 rule offenders}
A5 – Owner Matrix (P5): {diff_key → owner, SLA, fix PR/link, status}
A6 – Cutover Guardrails: kill-switches, rate caps, rollback triggers mapped to rule IDs
30-Minute Readiness Checklist (Printable)
P3 maps complete for 100 % of products in scope
Limit / entitlement rules source-controlled with version tags
First-Hour Ledger produced and signed by Finance + Ops
Diff Pack triaged — 0 critical, ≤ 3 medium with guardrails
P5 Owner Matrix populated; no orphaned diffs
Rollback conditions tied to specific rule IDs
Monitoring panels show rule-level counters, not only infra metrics
What Changes on the Ground (By Role)
CIO / CTO: Your green light is the ledger, not the runbook.
Chief Architect: Treat rules like code — diffable, testable, rollback-able.
Finance Controller: Co-own the First-Hour Ledger; pre-approve GL diffs.
Ops Lead: Drill kill-switches by rule group, not whole system.Program Manager: Tie every milestone to A1–A6 deliverables, not decks.
From Evidence to Culture
Once a team runs its first ledger-based simulation, the tone of readiness meetings changes.
People stop defending slides and start examining rules.
Executives move from “Are we ready?” to “Show me the ledger.”
That cultural shift — from presentation to proof — is what makes an enterprise coherent through change.
Turning Insight into Action
ICMG’s Enterprise Anatomy™ Diagnostics operationalize this shift:they link P1–P6 strategy, process, and logic into measurable, accountable artifacts that survive cutover and audit alike.
👉 Run your First-Hour Ledger Simulation before your next release.
👉 Join your industry’s Enterprise Anatomy™ Forum to see how others anchor go-lives in evidence, not decks.
When every owner signs their evidence, the enterprise stops collapsing at its own success.That’s the quiet power of being evidence-driven.
