Constraint-first management

Constraint-first management is the discipline of identifying the one operating system currently limiting the agency’s other systems, concentrating the next improvement cycle on that constraint, and verifying movement before starting anything else. It replaces six simultaneous half-finished initiatives with one finished one.

Why not improve everything at once?

Because improvement capacity is the scarcest resource in a working agency. Owners and teams have limited attention, and six parallel projects mean six things at 15% done — none of which changes the agency’s actual ceiling. A constraint, by definition, sets the limit for the whole. Capacity spent anywhere else produces activity, not throughput.

The pattern shows up concretely: an agency drowning in service exceptions does not benefit from a new lead vendor — the new volume converts to more exceptions. An agency whose owner makes every decision does not benefit from another hire — the new person queues behind everyone else waiting for answers.

How the constraint is found

Score all six systems honestly — including Unknown where evidence is missing — then ask which weak system most explains the symptoms in the others. Two rules of thumb:

• A system scored Unknown can be the constraint precisely because it is unmeasured — you cannot manage a limit you cannot see.
• When two systems tie, prefer the one upstream: focus problems create leverage problems; leverage problems create bench problems.

What happens next

One constraint becomes one accountable commitment: one operating page or tool, one metric, one owner, one review date. The decision loop closes on the review date — the metric either moved or it didn’t, and either answer is progress, because both narrow the next action.