Cycle Time

Running Cycle Time as a Daily Shop Floor Discipline

Cycle time sets the ceiling on everything a line can earn. This playbook covers how to time it honestly, what good looks like, and the daily cadence that keeps seconds from leaking away.

Cycle time is the ceiling on revenue for every asset you own. A line running a 45 second cycle produces 80 units per hour at best, and every 2 seconds of creep cuts capacity by about 4.4 percent. On a line selling output at 12 dollars a unit across two shifts, that 2 seconds is roughly 67 units and 800 dollars a day, or about 190,000 dollars a year. Plants that treat cycle time as a quoted constant instead of a managed number routinely give back 10 to 15 percent of capacity without a single machine breaking down.

Measure it with a repeatable method, not a guess. Time 25 to 30 consecutive cycles at the station, breakpoint to breakpoint, and record every one. Use the lowest repeatable time, typically the mode or the 20th percentile, as your standard, because the average absorbs minor stops and hides true capability. Then run the numbers: a 42 second standard against 27,000 seconds of net shift time supports 642 units. The Cycle Time calculator will cross-check cycle, takt, throughput, and utilization from those same inputs in seconds, so timing discipline is the only hard part.

Benchmark cycle time two ways. First, cycle-to-takt ratio: healthy lines run planned cycle at 85 to 95 percent of takt, leaving 5 to 15 percent buffer for variation. Above 95 percent you miss schedule on any hiccup; below 80 percent you are overstaffed or overbuilt. Second, variation: a coefficient of variation under 10 percent is solid for machine-paced work, while manual assembly commonly runs 15 to 25 percent until standard work tightens it. A station averaging 42 seconds with a standard deviation of 9 is a bigger problem than one averaging 46 with a deviation of 3.

The levers that actually move seconds are unglamorous. Move setup work external so the machine keeps cycling, which alone recovers 20 to 50 percent of changeover time. Put materials at point of use; every 3 meters of walking costs roughly 3 seconds per cycle. Fix machine parameters, since feeds and speeds set conservatively at launch often hide 5 to 10 percent of cycle. Split or rebalance work elements longer than takt. Attack the top three variation causes from your cycle study before chasing average reduction, because a stable 44 beats an erratic 41 for scheduling every time.

The common failure modes repeat in every plant. Timing the best operator on their best day, then wondering why the line hits 80 percent of standard. Averaging downtime and minor stops into the cycle number, which inflates standards by 10 to 20 percent and corrupts your OEE math. Quoting machine nameplate cycle from the OEM manual instead of demonstrated cycle. And re-timing stations only when an engineer has spare time, so standards drift 2 to 3 seconds a year uncorrected. Each of these puts a wrong number into scheduling, quoting, and staffing simultaneously.

Run cycle time on a cadence. Hourly: operators log actual versus planned counts on an hour-by-hour board, and any miss of 5 percent or more gets a reason code on the spot. Daily: the tier meeting reviews yesterday's top two reason codes in under 10 minutes and assigns one countermeasure. Weekly: run a fresh 30-cycle study on the constraint station and compare the distribution, not just the mean, to last week. Monthly: re-validate standards on every station that changed tooling, staffing, or method, which is usually 10 to 20 percent of the line.

World-class looks like this: cycle variation under 5 percent at every station, planned cycle within 90 to 95 percent of takt, standards re-validated monthly, and operators who can quote their station's cycle and its top loss without looking. Plants at that level respond to a 2 second drift within a shift, not a quarter. Getting there typically takes 6 to 12 months of the cadence above, and the payback shows up as 8 to 12 percent more output from the same assets, which is capacity you did not have to buy.

Published 2026-07-02.