Replacing a 30-year-old core banking system without a cutover weekend

The situation

The bank’s core ran on a mainframe platform its vendor would stop supporting within three years. Two earlier replacement attempts had stalled at the business case. The board wanted a plan it could trust, not another vendor promise.

What we did

1. Mapped every product, interface and manual workaround on the old core, about 1,400 in total, before speaking to a single vendor.
2. Designed a staged migration by customer segment, so each wave could be rolled back on its own.
3. Built a reconciliation layer that compared the old and new ledgers every night for six months.
4. Reported to the board monthly on risk, not on percentage complete.

What changed

• 2.1 million accounts moved across four waves
• No unplanned outage at any cutover
• 23% lower cost to serve in the first year

The lesson

Staging a migration costs more on paper and far less in practice. The board wasn’t buying speed. It was buying the ability to roll back.