Cycle time: measuring how long the work really takes at each step

Separating machine cycle from operator cycle

For any semi-automatic step there are two clocks: the machine running, and the person loading, unloading, checking and setting. Only one of them limits output, and which one it is determines the improvement worth funding. If the operator finishes well before the machine, adding a second machine to the same operator may be feasible; if the machine waits for the operator, tooling or fixture changes matter more than machine speed. Record the two separately from the start. Plants that log a single combined figure lose the ability to answer staffing questions and end up buying capacity they already have.

Getting an honest measurement rather than a flattering one

Direct observation changes behaviour, and observing a single skilled operator on a good day produces a figure the plant cannot hold. Take observations across shifts and operators, include the cycles that go wrong, and state clearly whether the figure includes routine interruptions. Where automatic counting is available, use recorded interval data over a long run instead, and reconcile it against the observed figure. The reconciliation itself is informative: a large gap between observed cycle and long-run recorded interval is precisely the minor stoppage loss that never appears in any breakdown report.

Variation matters more than the average

A step averaging a given duration with tight consistency behaves completely differently from one with the same average and a long tail of slow cycles. The tail is what fills the buffer in front of the next station, triggers stoppages and produces the shifts that inexplicably run short. Record the spread and look at what causes the slow cycles: material presentation, a tool nearing the end of its life, a fixture that needs persuading, a variant that carries extra content. Treating the tail as noise means planning against a duration the process only achieves about half the time.

Keeping cycle time and lead time apart in arguments

Cycle time is a property of a step; lead time is how long an order spends in the plant, most of which is queue and wait rather than work. Confusing the two produces a familiar error: halving a machine cycle in the expectation of halving delivery time, then finding that delivery barely moves because the part spent most of its stay waiting. Before funding a cycle improvement, establish what proportion of elapsed time is actually processing at that step. If it is small, the money belongs in flow, batching or queue control instead.

Keeping the recorded figures from going stale

Standard times decay silently. Tools get improved, fixtures get modified, materials change supplier, operators get better, and nobody updates the routing because no process requires it. Set a review trigger — engineering change, new material, sustained variance between planned and recorded run time — and give a named engineer the job. Sample the high-volume items regularly and compare planned against recorded duration. Where the plant consistently beats its standard, quoting is losing work it could win; where it consistently misses, every promise made from those routings is optimistic. Both errors are expensive and both are quietly self-perpetuating.

Frequently asked questions

How many observations do we need before trusting a cycle time?

Enough to see the process misbehave at least a few times, which means observing across operators and shifts rather than collecting many consecutive cycles from one person. Stop when additional observations stop changing the spread rather than when a count is reached. For a highly variable manual step you will need considerably more observation than for a machine-paced one, and if the spread stays wide the useful output is a list of causes, not a tighter average.

Should cycle time include inspection and paperwork?

Include anything the operator must do before the next unit can be started, because that is what limits the step. In-cycle checks, recording a reading, applying a label and confirming completion all consume the same clock. Record them as separate elements so they can be analysed later, but do not exclude them from the total. Excluding them is a common reason a line that meets its standard on paper falls behind in practice.

Why is our recorded cycle time better than our actual output rate?

Because output over a shift includes everything the cycle measurement excluded: minor stoppages, waiting for material, quality holds, adjustments, breaks and the slow cycles at the start and end of a run. The gap between the two is not an error, it is a measurement of loss, and it is usually the largest available capacity opportunity in the plant. Quantify it before spending on faster equipment.