At 8:15, a café has a long queue. The owner adds another cashier. Orders move faster, yet finished drinks do not. The attractive fix improved an activity; it did not improve the result customers were waiting for.
The missing question is simple: what step limits the flow of completed drinks?
See the whole system
Imagine demand for 100 drinks per hour. The order counter can process 120. A single espresso station can finish 60. Handoff can manage 90. Under these simplified assumptions, the café can complete at most 60 drinks per hour because the espresso station sets the ceiling.
Illustrative steady capacities. One drink per order; no hidden second constraint.
A faster cashier may still help service quality or cost. It simply cannot raise completed drinks above the espresso station's limit while that station remains the binding constraint. Raising the station's capacity to 80 could lift completed drinks to at most 80, assuming demand and the other steps hold.
Move the limit
Change one capacity. Watch what happens to completed drinks.
Espresso is the current limit. A faster order counter alone will not increase completed drinks.
Illustrative model: one drink per order, steady hourly rates and no hidden constraint. Output is the smallest of demand and the three capacities; it is not a profit forecast. Each example starts from the original values.
Five moves, in order
The Theory of Constraints gives managers a sequence for focusing improvement. First, identify the current constraint. Then exploit it: remove avoidable idle time and waste. Subordinate the other steps so they protect its useful time. Elevate its capacity when the extra output justifies the cost. Finally, repeat: once one limit moves, a different one may matter most.
- Identify: measure completed output and find the binding step.
- Exploit: keep the espresso station ready with cups, beans and maintenance.
- Subordinate: pace orders and handoff around the station's real capacity.
- Elevate: consider a second machine or shift only if the economics support it.
- Repeat: test whether handoff, demand or something else becomes the new limit.
This is a sequence for investigation, not a claim that the same machine will always be the problem. A queue is a clue; completed output, service rates and a small test make the diagnosis stronger.
The software version
The same logic can appear in a fictional software business. Suppose 800 qualified trials arrive each month, but an assisted onboarding team can complete only 200 setups. If 60% of completed setups become paying customers, that yields about 120 new customers. Doubling traffic could double trials while new customers stay near 120 if onboarding capacity and the other rates remain unchanged.
That example is about capacity. A conversion percentage alone does not create a hard ceiling. Before buying more clicks, this company should check what limits completed setups and whether improving that step is economically worthwhile.
A test to run this week
- Write down the result you want to improve: completed orders, paid activations, profit or another explicit measure.
- Map the few steps that produce it and estimate each step's effective capacity.
- Find where work waits, then check that diagnosis against completed output rather than queue length alone.
- Run one small change at the suspected constraint. Did the chosen result move?
If the answer is no, investigate what assumption failed before scaling the solution.
Sources and editorial notes
The five focusing steps and the concept of a system constraint draw on Goldratt Research Labs and the Theory of Constraints Institute. The café and SaaS calculations are original illustrative models for this episode. Their assumptions are stated above; no real company data is used.
When the companion video is approved and published, this page will link to it and record any later corrections.