Why the programme had stalled
Public-sector digitisation rarely fails on technology. It fails on the combination of shifting requirements, a vendor with no incentive to report bad news, and a service that cannot be switched off while it is replaced.
All three were present. The requirements moved because the policy underneath them moved — legitimately. The incumbent vendor graded its own progress. And citizens were still submitting paper, which meant any transition had to run both processes at once.
Phasing as a risk control, not a schedule
The temptation in a stalled programme is a decisive relaunch: new platform, fixed date, everything at once. In a public service that is the highest-risk option available, because the failure is visible to citizens and politically expensive.
Instead the paper process was retired in phases, each one small enough that the old route stayed available underneath it. That is slower on paper and considerably faster in practice, because no phase needed the heroics that a big-bang cutover requires.
Each phase also produced an auditable record, which mattered independently: a public body has to demonstrate not only that a case was handled, but that it was handled correctly and in order.
The hard part
Reporting honestly on a programme that had already been reported on optimistically. The first genuine status assessment was worse than the one it replaced, and delivering it was the precondition for everything that followed.
A delivery cadence that surfaces problems early only works if the first problem surfaced does not get the messenger removed. That is a governance question more than a project-management one, and it has to be settled before the first phase ships.
What transfers to other public-sector programmes
- Any vendor reporting on its own progress will report optimistically. Independent verification is not distrust, it is basic instrumentation.
- Phase around what can be reversed, not around what is convenient to build.
- Auditability is a requirement, not a feature — retrofitting it costs several times more than designing it in.
- The cadence you leave behind matters more than the platform you leave behind.