Shared work runs at the pace of its slowest interface, not at the pace of its fastest team.
Why it matters when the plan changes
Organisations routinely accelerate one part and assume the rest will follow. A product team reorganised for speed still depends on legal review, procurement and a finance cycle that were designed for a different rhythm. The acceleration is real inside the unit and invisible at the boundary, where the work actually has to cross, and where cross-unit commitments are already the least reliable part of execution: Sull, Homkes and Sull found in Harvard Business Review that only 9% of managers can rely on colleagues in other functions all the time.
The tension is that deliberately running at two speeds is often the right answer and is rarely designed. Where the interfaces between fast and slow are named and buffered, it works. Where the divide is simply allowed to emerge, coordination cost compounds with every added party, as Brooks's law describes, and the fast unit spends its capacity waiting and escalating, concluding the rest of the organisation is obstructive.
In practice
A newly formed digital unit is built to ship monthly. Every release needs a security review that runs on a quarterly committee. The unit's internal velocity triples and its delivery to customers does not change, and the review committee has no idea it has become the constraint.
Evidence
Cross-unit commitments are the least reliable part of execution, which is where a pace difference is actually paid for.
Donald Sull, Rebecca Homkes and Charles Sull, Why Strategy Execution Unravels and What to Do About It, Harvard Business Review (2015)Coordination cost rises with every party involved, so a mismatch at an interface compounds rather than averaging out.
Brooks's law, after Frederick P. Brooks, The Mythical Man-Month (1975)
What it cannot tell you
The pace divide identifies where a mismatch exists but does not tell you whether the slower side should speed up, the faster side should slow down, or the interface should change. It says nothing about whether a given control's cadence is proportionate to its risk, which is a separate design judgement.
Questions
Between a unit that has been deliberately accelerated and the functions it depends on: legal, security, procurement, finance and any centralised review. Sull, Homkes and Sull's Harvard Business Review research on execution found only 9% of managers can rely on colleagues in other functions and units all the time.
Yes. It works when interfaces between fast and slow are named and designed, with delegated authority or standing approvals at the boundary. Left undesigned, coordination cost rises with each added party, as Brooks's law, from Frederick P. Brooks's 1975 The Mythical Man-Month, describes, and the boundary becomes the actual constraint.
The faster unit, which experiences it as obstruction, and it usually escalates rather than diagnosing. The slower function is typically unaware it has become a constraint, because from inside its process nothing has changed and its own service levels are being met.
Changing the interface rather than either unit's speed: delegated authority for defined cases, a standing approval, a faster path for a category of decision, or resequencing so the slow step is not on the critical path. All are cheaper than accelerating a function.
No. Controls exist for reasons, and a review cycle that looks obstructive may be protecting something real. The question is whether the control's cadence still matches its risk, which is a design question rather than an argument about who is holding whom up.