Roadmapping

Definition

An architecture roadmap sequences change over time. Its quality is judged not by the destination but by whether each step is independently valuable and safe to stop after.

The property that matters most

Every step delivers something. A roadmap whose value arrives at the end is a project that will be cancelled before it gets there — and cancellation is the normal outcome for multi-year architecture programmes, because priorities change, sponsors move and budgets are reallocated.

Sequencing for incremental value is therefore not a preference; it is the risk mitigation that makes the programme survivable.

This is the same reasoning behind strangler-fig migration: value lands incrementally, each slice is validated in production, and the work can be paused in a stable state.

How to sequence

1. Start with what unblocks the most. A foundational capability many things depend on — identity, a deployment pipeline, a data platform — pays back across everything after it.

2. Then the highest pain-to-effort ratio. Early visible wins buy credibility for the harder work later, and credibility is the currency the programme runs on.

3. Respect real dependencies and question stated ones. Many claimed dependencies are sequencing preferences rather than genuine constraints, and challenging them frequently unlocks parallelism.

4. Include decommissioning as explicit work. Otherwise you run both old and new indefinitely, which is strictly worse than either alone — the characteristic failure of migration programmes.

What to put on it

Outcomes and capabilities, with rough time horizons — now, next, later — rather than precise dates for work eighteen months away, which are fiction and are treated as commitments.

Include what is not being done, and why. That prevents the roadmap being read as a complete plan and makes the deprioritisation explicit rather than assumed.

Keeping it honest

• Revisit quarterly. A roadmap not revised in a year is describing a world that has changed.
• State the assumptions, so it is clear which changes would invalidate the sequence.
• Track "what did we actually do against it", which is the only way anyone learns to sequence better.

Failure scenarios

• Value only at the end, so cancellation loses everything.
• Precise dates far out, treated as commitments and then missed.
• No decommissioning, so the estate only grows.
• A destination with no path, which is a vision statement rather than a roadmap.
• Never revised, so it describes last year's priorities.

Interview question

"How would you sequence a two-year modernisation so that cancelling it after six months still leaves the organisation better off?"