Contractual delivery needs governance: milestones, gates, budgets, defined acceptance. Technical work needs adaptation: findings emerge, priorities shift, an inspection result changes next week's plan.
Most engineering organisations already run both. The problem is rarely methodology. It is that governance lives in one system and the actual work lives in another, so the two are reconciled by hand every reporting cycle.
Why the waterfall-versus-agile debate misses the point
The debate assumes a choice at project level. In engineering services the split runs the other way: vertically. Commitments to the client are fixed at the top: scope, milestone dates, deliverable acceptance. The path to producing each deliverable is not fixed, and pretending otherwise produces plans that are obsolete within a fortnight.
A criticality assessment might reorder itself three times as equipment history arrives. The date it is due to the client does not move. Both statements are true simultaneously.
What must stay controlled
Anything the client can hold you to. Contracted scope and its exclusions. Milestone and acceptance dates. The commercial baseline and approved variations. Deliverable revision status and sign-off authority. These belong at portfolio level, under change control, visible to whoever answers for the contract.
What teams should be free to adapt
Everything that does not alter a commitment. Sequencing within a stage. Who picks up which analysis this week. How a study is decomposed into working sessions. Where review effort concentrates as findings develop.
When teams have to raise a change request to reorder their own week, they stop using the system and revert to a private spreadsheet. At that point governance reports become fiction, because they describe a plan nobody is following.
One record, two views
The resolution is not two systems kept in sync. It is one record read at two altitudes. A delivery lead sees the working items for the current stage. A portfolio manager sees gate status, variance against baseline and margin position. Neither view is a report compiled from the other; both are the same data.
An asset integrity programme runs across four platforms with quarterly gates and a fixed final acceptance date. Governance is stage-gated because the client's audit cycle requires it.
Inside stage two, the team works iteratively. Corrosion loop reviews are re-prioritised weekly as inspection data returns, and two loops are found to need a scope that was not anticipated.
Because the working items and the contracted baseline sit in one record, the additional loops surface immediately as variance rather than at the gate review six weeks later, when the recovery options would have been considerably worse.
What changes in practice
Governance and delivery read the same record
No reconciliation cycle between a portfolio view and a working plan.
Deviations surface in-stage, not at the gate
Recovery options remain open while they are still inexpensive.
Teams work in the system rather than around it
Day-to-day sequencing does not require a change request.
ProjectFlow holds the contracted baseline under change control while teams sequence their own work beneath it: one record, read at whichever altitude the question requires.
Explore ProjectFlow →