Improve the constraint, not the loudest complaint
When work feels busy everywhere, the best improvement is rarely a general push to work harder. Find the point where important work waits, returns, loses quality or creates demand downstream.
Choose one important outcome, map the few steps that create it and gather evidence from the people doing the work. Look for queue time, hand-offs, repeated decisions, missing information, rework and interruption. Test whether removing one constraint improves the whole outcome before expanding the change.
Trace one real piece of work
- Name the outcome, recipient and point at which it is complete.
- Follow recent examples from request to completion, including waiting, clarification, correction and escalation.
- Record duration separately from active effort; a five-minute task can wait five days.
- Ask the people doing the work which exception, hand-off or missing decision makes the normal route fail.
- Choose the smallest constraint with a credible link to the outcome and write a test.
Distinguish the signal from the symptom
| Visible problem | Possible underlying constraint | Useful check |
|---|---|---|
| Backlog grows | Unclear priority, variable demand, approval delay or rework. | Age work items and group reasons for waiting. |
| People are interrupted | No agreed response route, incomplete information or weak self-service. | Sample interruptions and what a better source would have answered. |
| Meetings multiply | Decisions, ownership or written updates are missing. | Review meeting purpose, decision record and actions. |
| Automation is requested | A repetitive step may be masking a poor or unstable process. | Measure variation, exceptions and correction before automating. |
Ask whether the bottleneck has moved
After a change, watch the next step and the user experience. A shorter local task can produce a larger queue, more defects or more support elsewhere. GOV.UK guidance on setting performance metrics recommends purpose-led measures, a baseline and context rather than isolated numbers.
Bottleneck FAQs
How many bottlenecks should we fix at once?
Start with one material constraint. Several simultaneous changes make it hard to learn what caused the result and can overload the people expected to adopt them.
What if the bottleneck is outside our team?
Make the dependency visible, show its effect on the shared outcome and work with the owner on a proportionate experiment. Avoid simply transferring pressure to another team.
Does a backlog always show poor productivity?
No. It can reflect demand, planned capacity, prioritisation, service level or a temporary event. Investigate age, consequence and flow before drawing a conclusion.
Continue your productivity decision
Use the next guide that fits the bottleneck, measurement, work design, communication, attention, automation, knowledge or review question still in front of you.