In most facilities, the time a unit spends being worked on is a minority of the time it spends in the building. The rest is waiting, and it is where the improvement is.
Measuring it
Every stage has a ready time and a start time. The difference is the wait.
Both are usually available from task and scan events, and the subtraction is the whole method.
Report by stage, as a distribution rather than an average.
Report by hour, which shows whether waiting is constant or concentrated.
Report the proportion of elapsed time that is waiting, which is the number that gets attention.
The typical picture
Waiting dominates. Handling is frequently a small fraction of elapsed time.
It concentrates at one or two stages, which is the constraint.
It varies enormously by hour, peaking after order release and before shift changes.
It is worse for exceptions, which wait for a decision rather than for capacity.
Why work waits
The next stage is busy, which is a capacity constraint.
The next stage is not staffed at that hour, which is a shift shaping problem and cheaper to fix.
Something is missing: stock, equipment, a document, a decision.
Batching. Work is deliberately accumulated before being processed, which trades cycle time for efficiency and is frequently not a conscious decision.
A dependency, where one task cannot start until another finishes.
Distinguish these before acting, because the remedies differ completely and capacity is the most expensive of them.
The fixes, cheapest first
Reshape shifts so staffing matches the demand curve. Costs nothing but rota negotiation and frequently removes a large share of waiting.
Reduce batch sizes, which shortens cycle time immediately.
Limit work release to what the constraint can absorb.
Remove the dependency where the sequence is habitual rather than necessary.
Cross-train, so capacity can move to where the queue is.
Add capacity at the constraint, which is the expensive answer and should be last.
The constraint moves
Fixing one bottleneck reveals the next, always.
Which is not a failure. It is how the improvement proceeds, and expecting it prevents the disappointment that follows the first fix.
Re-measure after every change, because the constraint will have moved and the previous analysis is now stale.
Keep the measurement running rather than treating it as a project, or the next constraint is discovered by complaint.
What to report
Waiting time by stage, weekly, as a proportion of elapsed time.
The current constraint, named.
The trend since the last change.
One chart, which is understood immediately by people who would ignore a productivity dashboard.
The queue chart
One chart, running permanently, that constitutes most of the operational control.
Queue length by stage, by hour, over the last two weeks.
The stage with the persistent queue is the constraint.
A queue appearing at a fixed hour is a release or staffing pattern, not a capacity problem.
A queue that never clears is a structural shortfall.
Post it where supervisors see it, and review it in the daily meeting.
Understood immediately by people who would ignore a productivity dashboard, which is why it is the chart to start with.
Batching as a hidden choice
Work is frequently accumulated before processing, and almost nobody decided to do it.
Find the batch points: where work waits for a quantity rather than for capacity.
Ask why the batch size is what it is. The answer is usually historical.
Batching trades cycle time for handling efficiency, which may be the right trade or may be a habit.
Halve one batch size and measure both sides: cycle time down, handling time up or unchanged.
Frequently the handling cost does not rise, which means the batching was pure delay.