The Home Office business case assumed 80% of the Police National Database's code could move to the cloud unchanged. In May 2025, after £35.1 million, the migration was scrapped: only 20% was reusable. The system is heading back to on-premises infrastructure, and its "long-term strategic direction" is, officially, "yet to be decided." That sentence is the post-mortem — not the cloud.
The assumption was the strategy
The programme was set up in 2021; actual work began in January 2024. Three years passed before anyone tested the one assumption everything else depended on. Lift-and-shift only works when portability is proven. Here it was priced in as fact — a planning convenience, not engineering analysis. When reality surfaced, the June 2025 deadline was gone, the support contract was expiring, and the supplier's contingency asked for another £26 million and 15–18 months. The cloud didn't fail. The premise did.
You pay twice for indecision
The arithmetic is damning: £35.1 million on a migration that won't ship, plus £20.3 million to stabilise the system back on-premises, plus roughly £24 million a year to run it for five to ten years — for a platform nobody has decided to modernise. That isn't a failed migration; it's a postponed decision dressed up as a programme.
My rule now: before funding the move, fund the truth. A portability assessment costs a fraction of a cancelled migration and tells you whether you're re-hosting or re-architecting. If your business case assumes 80% reuse, treat it as fiction until an engineer proves otherwise.