Every 3½ minutes, a car reaches Tomás. He is drowning.
Three stations down, another worker stands idle on the same line, at the same pace. A Yamazumi chart makes the imbalance impossible to ignore.
231 seconds of work.
| Work element | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | LRT |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Align rear drive unit | 60 | 62 | 58 | 61 | 60 | 63 | 60 | 59 | 61 | 62 | 60 |
| Torque 8 rear bolts | 47 | 49 | 48 | 47 | 50 | 47 | 48 | 51 | 49 | 48 | 47 |
| Retrieve connector · FL rack | 22 | 19 | 24 | 19 | 21 | 23 | 19 | 20 | 25 | 22 | 19 |
| Connect HV cable | 39 | 41 | 40 | 42 | 39 | 38 | 40 | 39 | 41 | 43 | 39 |
| Verify thermal sensor | 24 | 25 | 26 | 24 | 27 | 24 | 25 | 26 | 24 | 28 | 24 |
| Reset torque tool + scan | 42 | 55 | 44 | 42 | 60 | 47 | 42 | 50 | 45 | 58 | 42 |
was 5 before waste removal · staffed today: 6
Your station is packed. The one beside you breathes easy.
Tomás has 210 seconds to align a drive unit, torque eight bolts, run a high-voltage cable, and verify a sensor. He is the best tech in the zone, and still not fast enough. His lead, Marcus, has known for months. He just couldn't prove it.
A clipboard, a stopwatch, ten cycles.
No spreadsheet. Dana times every distinct element Tomás performs, across ten cycles, because no two run the same. Engineered standards say what a task should take; the stopwatch at the gemba says what it does. Capture it on a Time Observation Sheet. Our free one does the timing math for you.
Not the fastest. The lowest time that repeats.
Don't grab the single quickest number: a 58 that happened once is a fluke you can't bank on. Take the lowest time the operator hits consistently. And never average. When an element swings wildly (rows 3 & 6), that fluctuation is the job fighting back: a part that won't seat, a missing component, a quality issue. Average it and you bury those problems inside the standard.
Color each element by where it touches the unit.
This is the move that makes a Yamazumi more than a bar chart. Color by unit zone: rear-right, rear-left, front-left. The goal isn't to pin someone to one zone. It's flow: start at the front and work back, and a moving line moves with you. When the colors jump around (front, back, front), that's walking, chasing the vehicle.
Takt: the pace demand sets.
435 productive minutes ÷ 124 vehicles = 210 seconds. Drop the line across the bar and the verdict is instant: Tomás stands at 231 s, 21 seconds over. This one job, on its own, cannot keep up.
Six operators. One skyline.
Every station in Zone 3, side by side, against the same takt line. Station 7 spills over; Stations 8 and 9 sit idle. The imbalance you could only feel is now a shape on a wall, and the total work content is 980 seconds across six people.
Waste isn't a block. It hides inside the work.
The reflex is to shove tasks from the full station to the empty one. Wrong first move. X-ray each element and the waste appears as a fraction within it: the walk buried in “retrieve connector,” the wait inside “reset tool.” Improve the element; don't relocate it.
180 seconds of the easy waste, gone.
Move the Front-Left rack so the walk vanishes. Fix the torque fault so the wait disappears. The spread-out waste (the walks, the tool waits) comes off first: about 180 seconds across the line. Work content drops from 980 to 800, and every bar sits under takt. No one was asked to hurry.
How many operators do you actually need?
Add up all the real work and divide by takt: Σ 800 s ÷ 210 s = 3.81, round up to 4. The real floor was hiding in plain sight: the line is staffed with six. That's two operators of opportunity, visible before a single change hits the floor.
Now, and only now, move work.
With the easy waste gone, redistribute the elements into jobs loaded to about 90% of takt, not 100%, so a jam or a quality check doesn't stop the line. But elements are indivisible: you can't pour 800 seconds into four even buckets. You land at five: four jobs near takt, and a fifth holding the leftover that wouldn't fit.
The line breathes and points at what's next.
Six lopsided stations became five: four loaded near takt, and a fifth holding the isolated leftover. Tomás reaches without stepping; nobody drowns, nobody waits. And that underloaded fifth isn't a failure. It's the next kaizen target, made obvious. Absorb its work into the others and you reach the theoretical four. Six → five today → four next.
And it gives everyone a single picture of reality. No more “I think” or “it feels like.” Here's what's actually happening. That shared picture is what turns a room of opinions into alignment, and a team that can go fix it together.
You've seen it. Now stack your line.
The free Yamazumi builder: drop in your elements, color them by zone, draw your takt line, and watch the theoretical operator count update as you rebalance. No spreadsheet, no login.
Frequently asked
- What is a Yamazumi chart?
- An operator (or work) balance chart that stacks each person's work content side by side against takt time, so you can see overburden, idle time, and waste across a whole line in one picture.
- How is it different from a cycle-time bar chart?
- A plain bar chart tells you who is overloaded. Coloring each element by unit zone tells you why: layout mismatches and walking that a raw cycle-time number hides entirely.
- How many operators do I actually need?
- Add up the real work content across the line and divide by takt time (Σ ÷ takt), then round up. That theoretical minimum is frequently one or two fewer than the line is staffed for.
- Why not load operators to 100% of takt?
- Because real shifts have jams, missing parts, and quality checks. Loading to about 90% leaves a buffer so the first small problem does not stop the line, while still avoiding paid-for idle time.