Everything is coupled, so every project is the whole project.

With the domain that can move without waiting for anything else. Everything being coupled is exactly why breadth fails — a programme that cannot be cut into pieces cannot be delivered in pieces. We look for the seam that already exists rather than the domain that matters most.

Yes, and it is a common shape. They tend to carry long-cycle platform work while we take what needs to move faster or sit closer to the business. The arrangement works when ownership of each interface is written down, and it fails when the boundary is assumed.

By starting it early and expecting it to take real time. Security review, data processing agreements, supplier onboarding and insurance evidence are all predictable, so we begin them alongside discovery. The delay that hurts is the one nobody scheduled.

Every estate has them, and they usually carry business rules that exist nowhere else. We recover and test that logic before anything is replaced, because losing an undocumented rule is the failure that surfaces months later in a reconciliation nobody can explain.

Yes, and we ask for them at the start so the constraints are known before anything is designed. Where a standard is genuinely blocking something the business needs, we will make that case with evidence and let you decide. Undocumented deviation is how estates become unmaintainable, so we would rather have the argument in the open than work around a rule quietly.

Against a measure the business already tracks, with the baseline taken before anything changes. Cycle time, cost to serve, error rates, manual effort removed. Taking that baseline is the step most often skipped, and it is why so many programmes cannot say afterwards whether they paid for themselves.