A private bank had signed a core-banking replacement twice before — and cancelled twice. The third program, which we led end to end, cut over in 26 months and decommissioned the legacy core.
The bank's 30-year-old core was two major versions behind vendor support, and every prior replacement attempt had collapsed under scope growth: each department's "essential" customizations recreating the legacy system inside the new one. Meanwhile the migration itself — forty years of client positions, some in instruments that no longer traded — was the real risk nobody had priced.
We inverted the usual governance: an adoption budget instead of a customization budget, with every deviation from the vendor standard requiring a partner-level business case. Data migration ran as a first-class workstream from month one, rehearsed with full-volume dress runs — eleven of them — before the cutover weekend. Decommissioning the legacy core was written into the program charter as the definition of done.
“Eleven rehearsals felt excessive in month six. On the cutover weekend it felt like the only sane way to do this.”
Facing something similar? Talk to the partner who ran this →