The business is a chain of flows. One stage currently controls how much valuable output the system can create. Find it, protect it, and do not let unrelated improvement work steal the operating signal.
Acquisition, conversion, retention, and capacity are candidate lanes. The constraint is the lane whose improvement would increase the next valuable output after the system’s existing capacity and quality are respected.
Use the four checkpoints as a weekly control-room sequence. The active readout states what the operator should prove before moving to the next checkpoint.
Map acquisition, conversion, retention, and capacity with the same period, scope, and definition of valuable output. Pick the lane with the strongest combination of economic impact, evidence, controllability, and learning speed.
flow map → comparable evidence → candidate constraintEvery lane creates a different operating signature. Record the signal, its economic effect, and the first action that can disprove the diagnosis.
Watch qualified opportunities, source fit, response survival, and the ability of downstream teams to use the input. More raw leads are not automatically more throughput.
Watch stage yield, queue age, next-step completion, proposal aging, and contribution per win. A conversion leak is a specific stalled decision, not a general sales mood.
Watch activation, time to first value, usage, support load, contraction, and cohort survival. A cancellation reason is a clue; the leading behavior is the signal.
Watch throughput, work in process, cycle time, queue age, rework, defects, and the utilization of the suspected scarce step. Protect the step before adding resources.
A constraint review should leave an operator with a testable statement, not a motivational slogan.
The five moves below keep the team from buying capacity before it has used the system it already owns.
State the constraint as an observable step and connect it to the valuable output it limits. Do not use a department label as a diagnosis.
Remove avoidable interruptions, rework, incomplete inputs, and low-value work from the scarce step. Recover output with existing capacity first.
Align upstream volume, sequencing, handoffs, and priorities to the pace and quality requirements of the constraint. Stop feeding work the system cannot use.
Add people, spend, tooling, or distribution only after the first three moves show the constraint still governs output. Define the economic and quality case.
Measure throughput and downstream quality. If the constraint moved, reset the board. If it did not, reopen the diagnosis instead of defending the story.
Because the limiting step controls the next valuable output. Improving a non-constraint may make a local metric look better while adding queue, inventory, rework, or complexity elsewhere. One focused attack creates a clearer causal signal.
Define qualified input, then compare the rate and quality of that input with the stage where it stalls. If the system can use more qualified opportunities but does not receive them, acquisition is a candidate. If qualified opportunities arrive and age or leak at a decision step, conversion is the candidate.
Adding people before measuring the queue and removing avoidable loss. New capacity can hide unclear standards, poor intake, rework, or a decision bottleneck without increasing durable throughput.
Yes, if the retained cohort is too small, too low-margin, or unable to expand enough to support the acquisition and delivery system. Read cohort size, contribution, contraction, support burden, and time to value together.
The candidate lane, exact failure step, baseline, valuable-output measure, leading indicator, economic guardrail, owner, next intervention, review date, and the evidence that would move the constraint or falsify the diagnosis.
Do not leave the review with four projects. Leave with one constrained step, one owner, one test, and one date to see whether the board moves.
Open Constraint Trace