Takt is a requirement, not a measurement
Takt time is how often you must finish one unit to keep pace with the customer. Divide the time you have available to produce by the number of units demanded in that same period, and the answer is the interval between completions the process has to hit. The German word Takt means beat or metre — the sense a conductor uses — and the metaphor is exact: takt sets the tempo, and every station on the line plays to it.
The critical distinction is that takt is set by the customer and by your working calendar, not by the process. Speeding up a machine does not change takt. Improving quality does not change takt. Only two things do: a change in demand, or a change in how much time you make available. That is what separates takt from cycle time, which measures what the process actually achieves, and from lead time, which measures how long one unit takes to travel the whole route.
Getting the comparison right is the whole point. If cycle time is longer than takt, you cannot meet demand and the shortfall is arithmetically certain. If cycle time is much shorter than takt, you have capacity you are not using, and running the line flat out converts that capacity into inventory rather than into revenue. Lean practice aims to run at or just inside takt, and to use the difference deliberately rather than by accident.
Building the numerator honestly
Most takt calculations go wrong in the numerator. Net available production time is the time the line is genuinely expected to run: shift length minus breaks, minus the start-up meeting, minus planned cleandown, minus scheduled changeover. It is not the paid shift and it is not the shift minus every stoppage that actually happened.
The reason unplanned downtime stays out is that takt is a target, not a forecast. If you deduct last week's breakdowns from available time, takt gets shorter, and the line is asked to run faster to compensate for its own unreliability — which bakes the loss into the standard instead of exposing it. Unplanned losses belong in OEE, where they are visible and someone owns fixing them.
Multiply net time per shift by the number of shifts in the period your demand figure covers. The two must match: daily demand against daily available time, weekly against weekly. Mixing them is the most common arithmetic error in the whole calculation, and it produces a takt that is wrong by exactly the shift count.
Demand should be levelled. Using a peak week's orders produces a takt that sizes the line for the peak and leaves it idle the rest of the year; using an annual average produces a takt that cannot cope with any real week. Most operations use a levelled demand over a planning horizon of a few weeks and hold a small finished-goods buffer for the residual variation.
Worked example: two shifts building 1,600 units a day
An assembly line runs two 480 minute shifts a day. Each shift has 60 minutes of breaks and 20 minutes of planned stoppages for the start-up meeting and cleandown. The customer needs 1,600 units a day. Total manual work content is 300 seconds per unit, and finished units go into containers of 20.
- Net time per shift. 480 − 60 − 20 = 400 minutes.
- Net available time for the day. 400 × 2 = 800 minutes, which is 800 × 60 = 48,000 seconds.
- Takt time. 48,000 ÷ 1,600 = 30.0 seconds per unit.
- Required rate. 3,600 ÷ 30 = 120 units per hour.
- Theoretical stations. 300 ÷ 30 = 10 stations, exactly, with no balancing loss in theory.
- Pitch. 30 × 20 = 600 seconds = 10 minutes. A full container should leave the line every ten minutes, and if one does not, you know within ten minutes rather than at the end of the shift.
Now bring in the actual cycle time. Suppose the line currently completes a unit every 32 seconds. That is 32 ÷ 30 − 1 = 6.67% slower than takt. Over the day it produces 48,000 ÷ 32 = 1,500 units, which is 100 short of the 1,600 required. The shortfall is not a matter of effort; it is arithmetic.
There are only three levers. Reduce the cycle time to 30 seconds or less — 10 stations at 30 seconds each requires the work content to be balanced within 30 s per station. Add available time: an extra 100 units at 32 seconds each needs 3,200 more seconds, about 53 minutes of overtime. Or reduce demand on this line by moving volume elsewhere. Notice what is not a lever: telling operators to work faster within a station whose work content already exceeds takt.
Finally, see how takt responds to demand. Hold available time at 48,000 seconds and let demand rise to 1,920 units: takt becomes 48,000 ÷ 1,920 = 25.0 seconds and the theoretical station count becomes 300 ÷ 25 = 12 stations. Takt and demand are inversely proportional, which is why a 20% demand increase always shortens takt by exactly the factor 1 ÷ 1.2 = 0.833.
Reading takt against cycle time
Start with the sign of the gap. A positive gap means cycle time exceeds takt and demand cannot be met from the time available — the calculator reports the shortfall in units so you can size the fix. A negative gap means there is slack, and the size of that slack is a decision, not a result: you can absorb demand variation with it, run a shorter shift, or move work onto the line from elsewhere.
Then look at the theoretical station count. Work content divided by takt gives the minimum number of stations if the work could be split perfectly. It never can, so round up and treat the difference as balancing loss. With 300 seconds of work content and a 27 second takt, 300 ÷ 27 = 11.11, so you need 12 stations carrying 12 × 27 − 300 = 24 seconds of idle time spread across them. That idle time is the price of an indivisible task, and reducing it is what line balancing is for.
Use pitch to make the beat visible. A takt of 30 seconds is too fast to manage by eye, but a container every 10 minutes is not. Pitch is the interval at which a supervisor can see whether the line is on plan, and it is why a pitch board beats a shift-end report — it turns a daily variance into a ten-minute one.
One caution about running faster than takt. Every unit produced ahead of the beat is inventory: it consumes material, space and cash, and it hides the problem that made the extra capacity available. Lean treats overproduction as the most serious of the wastes precisely because it conceals the other six. If the gap is comfortably negative, the right response is usually to reduce available time or move work in, not to run flat out.
Takt time by demand for a two-shift day of 48,000 seconds
| Daily demand (units) | Takt time (s) | Required rate (units/hr) | Theoretical stations | Stations needed |
|---|---|---|---|---|
| 800 | 60.00 | 60.00 | 5.00 | 5 |
| 1,000 | 48.00 | 75.00 | 6.25 | 7 |
| 1,200 | 40.00 | 90.00 | 7.50 | 8 |
| 1,400 | 34.29 | 105.00 | 8.75 | 9 |
| 1,600 | 30.00 | 120.00 | 10.00 | 10 |
| 1,800 | 26.67 | 135.00 | 11.25 | 12 |
| 2,000 | 24.00 | 150.00 | 12.50 | 13 |
| 2,400 | 20.00 | 180.00 | 15.00 | 15 |
Takt is 48,000 ÷ demand and the required rate is 3,600 ÷ takt, which equals demand ÷ 13.333 hours of net running time. Where the theoretical station count is a whole number the line can balance with no idle time in principle; everywhere else the gap between the last two columns is balancing loss waiting to be designed out.
Mistakes that produce a wrong takt
- Mismatching the periods. Daily available time against weekly demand is wrong by the number of days. Both sides of the division must cover the same span.
- Deducting unplanned downtime from available time. That makes the line run faster to cover its own failures and hides the loss. Unplanned stoppages belong in OEE.
- Using paid shift time. Breaks, meetings and cleandown are not production time, and including them shortens takt to a beat the line was never given the time to hit.
- Using peak demand. Sizing to the peak leaves the line idle most of the year; sizing to the annual mean fails every busy week. Level the demand and buffer the residual.
- Treating takt as a process measurement. Takt does not improve when the line improves. Only demand and available time move it.
- Rounding the station count down. A theoretical 11.11 stations means 12 real ones. Rounding down silently assumes work can be split at any point.
- Ignoring the constraint operation. Actual cycle time should be measured where the line is slowest, not averaged across stations.
Takt, cycle time, lead time and pitch
- Takt time
- Net available production time divided by customer demand. A requirement set outside the process.
- Cycle time
- The interval at which the process actually completes units, measured at the constraint. A property of the process.
- Lead time
- The elapsed time for one unit to travel from order or raw material to completion, including all queueing. Usually far longer than cycle time × stations.
- Pitch
- Takt multiplied by the container quantity — the interval at which a full container should leave the line, and the natural frequency for checking plan against actual.
- Balancing loss
- The idle time created when indivisible tasks cannot be split evenly across stations: stations × takt minus total work content.
Where takt fits with the rest of the operating system
Takt is the demand end of the system, and it pairs with the capability end. OEE tells you what fraction of your planned production time actually produces good parts; takt tells you what rate that time has to deliver. Read together, they answer a question neither answers alone: if OEE is 75% and cycle time equals takt exactly, the line will miss demand roughly a quarter of the time, and either takt needs slack designed into it or the losses need removing.
Quality feeds the same equation from a third direction. Every unit scrapped has to be replaced within the same takt, so a process with poor capability effectively faces a demand higher than the customer's, by the reciprocal of its first-pass yield.
Downstream, pitch is what connects takt to material replenishment. A container leaving every ten minutes sets the kanban loop's timing, and the quantity in that loop is a reorder point problem: demand during replenishment lead time, plus safety stock for the variation you did not level out. Upstream, batching decisions that fight against takt — long runs to amortise a changeover — are the classic economic order quantity trade-off, and reducing changeover time is what lets the two stop fighting.
Finally, takt is not only for repetitive assembly. Applied to a hospital ward, a service desk or a claims process, the arithmetic is identical: available working time divided by cases to be handled. Where the work content varies widely per unit, takt is better used as a planning rhythm than as a station beat, and the calculator will say so when your takt exceeds an hour.