Iterate around business capabilities rather than phases. Each cycle picks one capability and runs it through the phases end to end.
Phases per cycle. Phases B through F inside a single capability slice. Phase A sets the capability map once, up front; Phases G and H stay continuous across all capability cycles.
Trade-offs. Gives the portfolio a clean way to sequence investment but risks losing cross-capability coherence if no one holds the seams. Needs a capability owner per slice who can make trade-offs without waiting for a steering board.
Good for: Portfolio-wide change, operating model redesigns, any programme where the board talks in capability language rather than phase language.
Baseline-first iteration
Describe the as-is fully before designing the to-be. Good for regulated or change-averse settings where the current picture must be defensible before anyone draws a target.
Eight to twelve weeks per cycle.
Target-first iteration
Sketch the to-be first, then back-fill the baseline only where it directly informs the change. Good for high-speed transformation with a clear strategic bet.
Four to eight weeks per cycle.
Capability-based iterationRecommended
Iterate around business capabilities rather than phases. Each cycle picks one capability and runs it through the phases end to end.
Six to ten weeks per capability cycle.
Architecture-domain iteration
Pick one architecture domain (business or data or application or technology) per cycle and go deep. Good for specialist depth when the team has the right senior people.
Four to six weeks per domain cycle.
Solution-driven iteration
Run ADM cycles alongside solution delivery sprints. The architecture stays one or two sprints ahead of the build team, no further.
Two to three weeks per cycle, sprint-aligned.
