How to Create a Yamazumi Chart

Steps to build a Yamazumi chart

1. 1

   Select the process and fix its boundaries

   Pick one line, cell or workflow — not a whole factory — and define exactly where each cycle starts and ends. Does it begin when the operator picks up the first component, or when the product arrives from the previous station? Does it end at inspection, or at transfer? The specific choice matters far less than applying the same one to every observation. Inconsistent boundaries are the single most common reason a study produces numbers that cannot be compared with each other.

2. 2

   Observe before you measure

   Watch several complete cycles without a stopwatch running. Documented work instructions and actual practice diverge in almost every plant, and the differences — an extra trip to the rack, a tool repositioned, a part checked twice — are usually where the recoverable time is. Note what surprises you. Engineers who measure before they understand the process end up with accurate times for the wrong elements.

3. 3

   Choose how you will collect the data

   A stopwatch is fast and cheap, and remains reasonable for a short, simple cycle — but the observer's reaction time enters every reading, the observation cannot be reviewed, and being timed changes how people work. Video removes all three limits: boundaries can be set precisely, a second engineer can check the split, and the recording remains as evidence when a figure is challenged months later. Data from machine monitoring systems is useful for cycle counts but rarely explains why a loss occurred, so it should be validated against direct observation.

4. 4

   Break the work into elements

   Split each station into small, individually timed activities instead of recording one figure per operator. Rather than 'assemble valve', record picking the body, positioning it, installing the seal, inserting the shaft, tightening fasteners, inspecting and transferring. A good element is clearly identifiable, repeatable, performed by one person and completed without interruption. Fine elements are what make rebalancing possible later — you can move a 6-second element between stations; you cannot move a 58-second lump.

5. 5

   Measure each element over several cycles

   Time every element across at least five cycles, and ten or more where the work visibly varies, then average. The purpose is not precision for its own sake — it is to stop one unusually fast or slow cycle from setting the target for a station. Measure under normal running conditions and exclude training, trials, breakdowns and material shortages. Where an element swings widely between operators, record the spread as well as the mean: that gap is usually missing standard work, and it will not show up in an average.

6. 6

   Classify every element

   Tag each element as value-added (physically transforms the product in a way the customer pays for), incidental work (necessary but adds no value — inspection, documentation, machine loading) or waste (walking, waiting, searching, double handling, rework). This step is skipped more often than any other and it is the one that makes the chart actionable. Without it you know how long a station takes; with it you know whether the answer is to move work, change the layout, or delete the activity.

7. 7

   Calculate takt time

   Takt is net available production time divided by customer demand for the same period. A shift with 450 productive minutes against demand for 500 units gives 27,000 ÷ 500 = 54 seconds. Note that it comes from demand, not from what the line currently achieves — takt is the requirement the chart is measured against, and it changes whenever demand or the shift pattern changes.

8. 8

   Assign elements to stations and total them

   Group the measured elements under the operator or station that performs them, and total each. A station running pick housing 6.0, position 8.5, install bearing 12.0, insert shaft 9.0, tighten bolts 15.5, inspect 6.5 and transfer 4.0 totals 61.5 seconds. Repeat for every station. At this point you already know which stations exceed takt — the chart is about to make it visible to everyone else.

9. 9

   Build the chart

   One vertical bar per station, one coloured segment per work element, segment height proportional to measured time, and a horizontal takt line across the whole chart. Use consistent colours for the three work classes so the value-added share is readable at a glance. The bar total is the station's cycle time; the takt line is the requirement. Everything the chart is for follows from the relationship between those two.

10. 10

    Validate before drawing conclusions

    Check the chart against reality before it leaves the office. Were enough cycles observed? Do the totals match what supervisors see on the floor? Are indirect tasks — replenishment, paperwork, tool changes — included or knowingly excluded? Does the sequence match what operators actually do? Show it to the operators whose work it describes; they will find omissions faster than any review, and their involvement is what makes the subsequent change stick.

11. 11

    Find the constraint and its cause

    The tallest bar over takt sets the pace of the line, but the useful question is what it is made of. If it is nearly all value-added, the station is genuinely full and work has to move. If a fifth of it is walking and searching, the answer is layout and material presentation, not reassignment. And check that the overload is labour at all — a machine cycle, an inspection queue or a material shortage will not be fixed by moving elements between operators.

12. 12

    Rebalance, update standard work, then verify

    Remove waste first, because that reduces total work content rather than relocating it; only then move elements to stations with capacity. Aim for every station under takt with some margin, not for identical bar heights — perfect numerical balance is neither achievable nor desirable when stations absorb variation or carry critical quality steps. Update standard work for every station you touched. Then rebuild the chart after implementation and compare like for like: without that second chart, an improvement is a claim rather than a result.

The takt line has to be drawn before the bars mean anything, and it comes from two numbers rather than from measurement — calculate takt time for your line.