Case study
Migrating a Global Bank’s Public Web Estate Without a Day of Downtime
Leading the phased migration of a multi-market banking web platform — 40+ country sites, regulatory sign-offs, and zero downtime at cutover.
- Client
- Global banking group
- Industry
- Banking & Financial Services
- Role
- Technical Project Manager
- Duration
- 14 months
Challenge
A public web estate spanning 40+ country sites sat on an end-of-life CMS. Every market had its own regulatory sign-off chain, content freeze windows and localisation stack — and the business mandate was zero customer-facing downtime during migration.
Approach
Phased market-by-market migration with a repeatable playbook: automated content audits, a parallel-run architecture allowing instant rollback, and a governance cadence that got regional compliance teams approving in parallel rather than in sequence.
Outcome
All markets migrated ahead of the platform’s end-of-support date with no unplanned downtime. The playbook cut per-market migration time from 6 weeks in the pilot to under 2 weeks by the final wave.
- 40+
- Country sites migrated
- 0 min
- Unplanned downtime
- 6wk → 2wk
- Per-market migration time
- 120+
- Compliance sign-offs coordinated
Where it started
End-of-life software in a bank is not a technical inconvenience — it is a compliance event. When the group’s web content platform was announced end-of-support, more than forty country websites, each with its own regulators, languages and legal disclaimers, needed a new home. The previous attempt at a “big bang” replatform had been cancelled after a year of planning; the appetite for risk was zero and the deadline was fixed.
Making the problem smaller
The first decision was the one everything else depended on: no big bang. We would migrate market by market, in waves, with the old and new platforms running in parallel and traffic switchable per market in minutes.
That single choice transformed the risk profile. A failed cutover in one market was a contained, reversible event — not a group-level incident. It also let us learn: the pilot market took six weeks and surfaced dozens of gaps in our content audit tooling; each subsequent wave banked those lessons into the playbook.
The governance unlock
The genuinely hard problem was not technical. Each market’s compliance and legal teams expected to review the migrated content sequentially — after the previous market finished — which multiplied out to a timeline far beyond the end-of-support date.
The unlock was redesigning the sign-off process itself. We produced a standard evidence pack per market — automated content-diff reports proving nothing had changed except the platform — and pre-briefed regional compliance leads as a group. Reviews became parallel and evidence-based rather than sequential and manual. Sign-off time per market dropped from weeks to days, and 120+ approvals landed without a single one blocking a wave.
What I would repeat
- Parallel-run with instant rollback as a precondition, not an optimisation. It bought us the political permission to move fast.
- A migration playbook treated as a product — versioned, improved every wave, owned by the team that used it.
- Evidence packs over meetings for regulated sign-offs. Compliance teams do not want reassurance; they want proof they can file.
The result
Every market moved before end-of-support, with zero minutes of unplanned customer-facing downtime. The final waves ran so smoothly that migrations stopped being a leadership agenda item — which, for a programme that began under cancelled-predecessor shadow, was the quietest possible definition of success.