Why most roadmaps decorate instead of govern
The standard transformation roadmap is a wall of coloured bars whose real function is reassurance: it proves a plan exists without committing anyone to a consequence. Dates slip, the bars stretch, and no decision is ever triggered, because the roadmap was never wired to any. The test of a roadmap is simple. If every phase can run late and over budget without forcing a decision about the next one, it is a picture of intent, and the organisation is funding intent.
The phased roadmap
| Phase |
Outcomes it must deliver |
Dependencies |
Decision gate question |
Funding released when |
| Phase 0: Baseline and design |
Verified current-state costs, target operating design, sequencing plan with owners |
Finance sign-off on the baseline; senior time actually diarised, not promised |
Is the business case still true at the verified baseline? |
Design funding only. Build funds stay locked until the gate answers yes |
| Phase 1: First working slice |
One process or unit running end to end in the new model, with measured results |
Data access agreed; the receiving team backfilled so it can absorb the change |
Did the slice deliver its forecast outcome at forecast cost? |
Phase 2 funds released against measured results, not against effort expended |
| Phase 2: Prove repeatability |
The slice replicated in a second and third unit, with a known cost per rollout |
Phase 1 lessons documented and absorbed; change capacity confirmed in receiving units |
Does the cost and time per unit support the full rollout case? |
Rollout tranche released only once the per-unit economics hold twice |
| Phase 3: Core migration |
The main volume moved, legacy dependence broken for critical processes |
Data quality remediated to agreed thresholds; cutover rehearsed, not merely planned |
Do rehearsal results justify go, or does the plan restructure? |
Migration tranche released against rehearsal evidence and a tested rollback |
| Phase 4: Decommission |
Legacy systems retired, contracts exited, run-cost reductions landed in named budgets |
Contract exit dates mapped; benefit owners' budgets adjusted in the plan, in writing |
Have the promised savings actually left the cost base? |
The final tranche pays for decommissioning, never for go-live celebrations |
| Phase 5: Embed and hand over |
Ownership moved from programme to line; capability to change the system without the integrator |
Knowledge transfer completed against a dated plan with named people |
Can the organisation run and adapt this without external help? |
Residual funds released at handover; anything unspent returns to the centre |
Making the gates able to say no
A gate is only real if three conditions hold. The chair of the gate review does not report to the programme. The funding is genuinely tranched, held by finance rather than pre-committed to vendors. And at least one plausible gate outcome is stop or restructure, written down in advance, so that saying no is a foreseen path rather than a crisis. A gate that has never stopped, shrunk or resequenced anything in the organisation's memory is a signing ceremony, and everyone attending knows it.
How roadmaps drift back into decoration
- Outcomes get restated as activities ("complete workstream three") so every phase can succeed without anything changing in the business.
- Gate dates move to match delivery rather than delivery being tested at the gate, which converts governance into scheduling.
- All the benefit is stacked into the final phases, making the early gates unfalsifiable and the programme unstoppable by design.
- The dependency column fills with other teams' names that were written in without asking them, and each one is discovered as a surprise later.
- The roadmap is re-versioned so often that no baseline survives to compare against, and slippage becomes structurally invisible.
When the sequencing itself needs outside challenge
Internal reviews test whether the roadmap is deliverable. The harder question is whether it is the right sequence at all: whether the first slice is genuinely provable, whether repeatability is being tested on the easiest units, and whether the integrator's preferred order serves your risk profile or its revenue profile. Senior operators who have run comparable programmes, including ones that stalled, read a roadmap in an afternoon and tell you which gate will be the one that gets fudged. Hearing that before approval is what a Digital Advisory brief is for.
