Takt time
Takt time (often shortened to takt) is the working time available over a period divided by the number of units that customers demand over the same period. It gives the interval at which a process must deliver one unit to keep up with demand: 465 minutes of work for 155 orders a day give a takt of 3 minutes. The team compares this pace with the cycle time measured at each step.
Goal
Takt time is the pace of demand expressed as time. According to the Lean Enterprise Institute (LEI), takt serves to match production to demand. In German, Takt is the musical bar. The LEI places its first industrial use in the German aircraft industry of the 1930s; Toyota made it general practice in the 1950s and 1960s.
The team measures the cycle time of each step and compares it with the takt. A slower step delays every unit that passes through it: the queue in front of it grows every day. A much faster step produces stock that nobody asked for. When the work cannot be stocked, the people at that step wait for the next piece of work.
The technique prepares decisions on staffing and on the split of work: how many people, which steps to combine, which step to shorten, when to lengthen the working day. Art Byrne places takt at the base of single-piece flow, standardised work and the pull system. The deliverable is a takt value with its basis: the available time, the demand, the period and the date of the next review. A spreadsheet row or a board at the workstation carries the deliverable.
Usage
When to use it
- Repetitive work on a comparable unit: orders, laboratory analyses, parts of the same family; takt sets the pace of each step.
- Designing a single-piece flow cell: allocate the steps across people so that no workload exceeds the takt.
- Writing standardised work: takt is the first element on the standardised work chart.
- The future-state map in value stream mapping: give each step the pace it must hold.
- A step suspected of holding up the flow: compare its measured cycle time with the takt to confirm the bottleneck.
- Demand that moves to a new level: recalculate staffing and the allocation of steps across people from the new takt.
- An administrative service with estimated demand: set a pace from history and see where work piles up.
- A team spread across several sites: takt is calculated remotely; cycle time is measured on each site or in the tool that timestamps the steps.
When not to use it
- Work with no family of comparable units: an average takt says nothing about a single case; follow the flow with a cumulative flow diagram.
- A commitment to a single deadline: the question concerns a date; plan with the critical path method.
Description
The calculation
Takt = available working time ÷ demand
The available time and the demand cover the same period.
The available time is the net time during which the team can work. Common practice starts from attendance over the period. It deducts breaks and planned downtime: the morning team meeting, scheduled maintenance. Breakdowns and unplanned absences stay in the available time. Deducting them would shorten the takt. The shorter takt would then hide the breakdown in the pace: nobody would treat it as a problem.
Demand is the number of units that customers request over the period. Ohno divides the month's quantity by the number of working days, then the day's working time by the resulting quantity. According to Michael Ballé, the demand on a service is never known in advance. The team estimates it from a few weeks of history.
The team chooses the period and the unit together. The LEI cites a pace of two new products a month: the takt is then two weeks. The formula applies to product development with a longer period.
The team's speed does not change the takt: a team twice as fast has the same takt.
Takt, cycle time and lead time
The team reads takt beside two durations that it measures on the process.
| Duration | What it says | How the team obtains it | What makes it vary | Order fulfilment centre |
|---|---|---|---|---|
| Takt | The interval at which the process must deliver one unit to keep up with demand | Calculation: available time ÷ demand | Demand, available time | 3.0 minutes per order |
| Cycle time | The time a step takes to process one unit | Measurement at the step, over several units | Method, staffing, tools | Packing: 2.8 minutes |
| Lead time | The time a unit takes to pass through the whole process | Measurement from entry to exit | Work in progress, waits between steps | Under a day: order received in the morning and handed to the carrier in the evening |
The LEI defines cycle time as the time needed to produce one part or complete a process. It comes from an actual measurement. Takt sets the pace that the process must hold. Cycle time states the pace that it holds.
Names vary from one source to another. In the English edition of Ohno's book, cycle time means the time allotted to one unit. Ohno calculates it from the quantity demanded: it is the LEI's takt. Several kanban tools call the time between the start of work on an item and its delivery the cycle time. You can read this duration on a cumulative flow diagram. The team writes its definition of cycle time at the top of the record. It measures cycle time by that definition before it compares it with the takt.
Takt and throughput
Throughput counts the units that the process delivers per period. With 465 minutes available and 155 orders a day, demand calls for one order every 3 minutes. A team that delivers 140 adds 15 orders to the work in progress every day. Over a stable period, Little's law links these quantities: average lead time equals average work in progress divided by average throughput. At constant throughput, more work in progress lengthens the lead time of every order. The cumulative flow diagram shows the gap between arrivals and deliveries week after week.
Running the calculation
- Choose the process and the unit. An order, a refund request, a building permit file. Pick a family of units that follow the same steps with comparable durations.
- Establish the available time. Start from the team's attendance over the period. Deduct breaks, the team meeting and planned downtime. Record each deduction so that someone else can redo the calculation.
- Estimate demand. Take the average of a few weeks of history for the same unit and the same period. Record known peaks separately: month-end, December, the end of the summer holidays.
- Divide. Express the takt in the unit that people read at the workstation: seconds, minutes, hours or days.
- Measure the cycle time of each step. Time several units on site with the people who do the work. When the tracking tool timestamps each step, its data gives a first measurement. Compare each step with the takt.
- Decide. The team shortens the step above the takt, duplicates it or assigns one more person to it. Minimum staffing equals the total work of one unit divided by the takt and rounded up to the next whole number. When every step stays far below the takt, the team combines steps and frees people for other work.
- Display and track. Display at the workstation the number of units expected each hour beside the number delivered. Byrne sees this tracking as a visual control of progress against demand: anyone reads the gap without calculating.
- Review on a fixed date. According to the LEI, Toyota reviews the takt of a process every month and adjusts it every ten days. Record the date of the next review with the value.
When to change the takt
Demand varies from day to day. Ballé advises against changing the takt at every variation. The team absorbs repeated gaps with stock, with the length of the working day or by adding or removing a shift. These levers leave the takt in place. Before it changes the takt, the team looks for the reason why demand moved. A known season, a new customer or a campaign has a lasting effect on demand: the team calculates a new takt and allocates the steps across people again. According to Ballé, each change of takt obliges the team to rebalance every workstation. The team therefore limits these changes. A single day does not justify a new takt.
- 155 orders a day: takt 3.0 min
- 186 orders a day: takt 2.5 min
- day-to-day gaps: orders waiting or length of the working day
- review on the first Monday of December
The LEI calls the takt multiplied by the pack-out quantity the pitch: at 3 minutes per order and 20 orders per trolley, the pitch is 60 minutes. Production levelling (heijunka) relies on this interval.
Takt in a service
An administrative service receives requests that it does not schedule. Even so, Ballé calculates a takt for such a service: opening time divided by average demand. The team delivers at that pace and compares the demand received with the takt every week. A lasting gap means that demand has changed. For Ballé, takt uncovers problems such as batch work.
Take a service that opens five days a week for 420 minutes a day. It receives 40 files a week: the takt is 2'100 ÷ 40 = 52.5 minutes. That is eight files a day. A team that processes its files in a batch on Friday delivers none from Monday to Thursday. The shortfall against the takt from Monday to Thursday shows the batch.
Pitfalls
Treating attendance as available time. With 510 minutes of attendance instead of 465 net minutes, 155 orders give a takt of 3.3 minutes instead of 3.0. A 3.1-minute step looks below the takt and yet builds a queue.
Confusing takt and cycle time. A duration derived from the team's speed is a cycle time. A hospital length of stay or the processing time of a file is a lead time. Takt comes from demand and available time, never from the team's speed.
Loading each step to the exact takt. A 3.0-minute step for a takt of 3.0 has no margin. One slightly longer unit creates a queue that the step never clears. The team keeps a gap between the highest workload and the takt. At baseline, packing at 2.8 minutes leaves only 0.2 minute against a takt of 3.0: an order with several items pushes packing over the takt.
Calculating one takt over dissimilar units. An order of one item and an order of forty items do not take the same time. A shared takt hides both. The team calculates one takt per family.
Reading takt as a pace imposed on people. Takt sets the pace of the process. A step above the takt signals a problem of method or of staffing. The team deals with it on the process.
AI considerations
An export from the tracking tool gives the entry date of each order, ticket or file. A language model or a script derives from it the demand per day, per week and per family. It spots the recurring peaks at month-end or in December. It calculates the takt for each demand scenario. When the tool records the start and the end of processing at each step, the same export gives a first estimate of cycle time.
The model cannot see what the export leaves out: the Monday team meeting, part-time hours, the carrier's cut-off time. The team sets the available time itself. Nor does the model know why demand moved. A campaign, a new customer and an import error draw the same curve. Only the team decides to change the takt.
Art. 26 of Ordinance 3 to the Employment Act (OLT 3) prohibits systems designed to monitor the behaviour of workers at their workstation. A takt analysis that ranks employees by speed would become one. The team removes names before the analysis and measures steps.
Examples
An online shop in French-speaking Switzerland calculates the takt of its order fulfilment centre. The unit is an order. The table calculates the takt for three levels of demand and sets the measured cycle time of each step against it.
| Line | July | Baseline | December |
|---|---|---|---|
| Attendance per day, 7:30 to 17:00 less one hour at lunch | 510 min | 510 min | 510 min |
| Breaks (2 × 15 min) and team meeting (15 min) | −45 min | −45 min | −45 min |
| Available time | 465 min | 465 min | 465 min |
| Demand per day | 93 orders | 155 orders | 186 orders |
| Takt | 5.0 min | 3.0 min | 2.5 min |
| Picking, 2.6 min | below takt | below takt | above |
| Checking, 1.2 min | below takt | below takt | below takt |
| Packing, 2.8 min | below takt | below takt | above |
| Labelling and handover to the carrier, 1.0 min | below takt | below takt | below takt |
| Total work per order | 7.6 min | 7.6 min | 7.6 min |
| Minimum staffing (total work ÷ takt) | 1.52: 2 people | 2.53: 3 people | 3.04: 4 people |
| Allocation | 2 people: A picks and checks (3.8 min); B packs and labels (3.8 min) | 4 people: one at each step. The minimum of 3 would combine two adjacent steps. Each pair exceeds 3.0 min: picking and checking 3.8; checking and packing 4.0; packing and labelling 3.8. | 5 people: one at each step plus a second at packing. Half an hour more per day (495 min) covers picking: 190 orders at 2.6 min. |
| Demand basis | July of the previous year | Average of the last 8 weeks | December of the previous year |
| Review | The first Monday of every month | ||
- July: takt 5.0 min (93 orders a day)
- Baseline: takt 3.0 min (155 orders a day)
- December: takt 2.5 min (186 orders a day)
The cycle times stay the same across the three columns. Only the takt moves. In December, the takt drops to 2.5 minutes: picking and packing exceed it. Half an hour more per day gives 495 minutes of work. At 2.6 minutes, picking then delivers 190 orders and covers the 186. At 2.8 minutes, packing delivers only 176. The takt does not change. The team assigns a second person to packing.
Visualisations
The table of the three durations separates takt from cycle time and lead time. The team calculates the first from demand. It measures the other two on the process. The demand chart sets six weeks of daily orders against the level that fixes the takt: day-to-day gaps leave the takt in place, the change of season changes it. Over one week, the service chart accumulates the files expected at takt and the files delivered in a Friday batch. The takt calculation table gives three levels of demand and one row per step. The cycle time chart sets the four steps of the example against the three takts.
Cost
| Dimension | Level | Note |
|---|---|---|
| Preparation | Low | Collect attendance, breaks, planned downtime and a few weeks of demand from the tracking tool. |
| Execution | Medium | The division takes a minute. Measuring the cycle time of each step requires observing several units on site. |
| Documentation | Low | The team reviews one spreadsheet row and one workstation board every month. |
Tools
A spreadsheet is enough for the calculation: one column per demand level, one row per step. It recalculates takt and staffing when demand changes. The export from the tracking tool (order, ticket or case management) gives the demand history. When the tool timestamps each step, the export also gives a first measurement of cycle times. A stopwatch and an observation sheet measure the steps that the tool does not see. At the workstation, a whiteboard carries the takt, the number of units expected each hour and the number delivered. A cumulative flow diagram tracks throughput against demand over several weeks.
Sources
- Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production, Productivity Press, 1988 (Japanese original, 1978): the time allotted to one unit, calculated from the month's quantity and the number of working days, under the name cycle time.
- Lean Enterprise Institute, Takt Time, Lean Lexicon: the definition of takt and its purpose, the example of two products a month, the origin of the word and of the practice, the monthly review at Toyota.
- Lean Enterprise Institute, Cycle Time, Lean Lexicon: cycle time taken from an actual measurement.
- Lean Enterprise Institute, Heijunka Box, Lean Lexicon: the pitch as the takt multiplied by the pack-out quantity.
- Art Byrne, Ask Art: Why Is Takt Time So Important in a Lean Turnaround?, Lean Enterprise Institute: takt at the base of flow, standardised work and the pull system; the visual control of progress against demand.
- Michael Ballé, How Often Should We Change Takt Time?, Lean Enterprise Institute: absorbing variations in demand without changing the takt; looking for the reason why demand moved.
- Michael Ballé, How Do You Apply Takt Time to Service Work?, Lean Enterprise Institute: demand estimated from history; takt as a way to uncover batch work.
- Ordinance 3 to the Employment Act (OLT 3), SR 822.113, art. 26: the prohibition of systems designed to monitor the behaviour of workers at their workstation.

