Toyota Kata

Thinking Tools6.1 Designing and Testing the SolutionAlso known as Improvement Kata, Coaching Kata

Developed by Mike Rother from the management routines he observed at Toyota (Toyota Kata, 2009).

Pairs a routine for working toward a target condition through small experiments with five standard coaching questions a manager asks daily to build the habit.

The five coaching questions at a packing station, and the answers a coach can accept

A shift lead is working the packing station toward a target of 45 seconds an order. At each of the learner's turns pick the answer; the tag says whether a coach should accept it or ask again. Illustrative.

Coach

Learner

Coach

Learner

Select a stage to read it, or play the sequence.Ask: the target. The coach asks at the station's board, every morning at the same time, the same five questions in the same order.Answer: the target. The coach has asked what the target condition is. Pick the shift lead's answer.Pick a move to see the reply.Coach: "Good. Now, where are you today?"Accept: a condition at the process, with a number and a date.Coach: "How would we know you got there?"Ask again: a wish, with nothing to measure.Ask: where now. Then the coach asks for the facts, and what was learned from the last step.Answer: where now. The coach has asked for the actual condition. Pick the answer.Pick a move to see the reply.Coach: "What made the slow ones slow?"Accept: measured at the station, with the spread as well as the average.Coach: "Let's go and time a few together."Ask again: an impression, and no number to compare with the target.Ask: obstacles. The coach asks what stands between the actual condition and the target, and which one the learner is taking on now.Answer: obstacles. The coach has asked about obstacles. Pick the answer.Pick a move to see the reply.Coach: "What's your next step on it?"Accept: one obstacle, seen at the process, with a size.Coach: "Which one are you working on today?"Ask again: a list too big for one step, and none of it chosen.Ask: next step. The coach asks for one experiment and a prediction, so the result can teach something.Answer: next step. The coach has asked for the next step and the expected result. Pick the answer.Pick a move to see the reply.Coach: "When can we see what happens?"Accept: small, testable today, with a prediction to check.Coach: "What could you try by tomorrow?"Ask again: a project, and nothing to learn from this week.Coach: "What exactly, and what do you expect?"Ask again: no step named and no prediction.Ask: when we see. The last question sets the next meeting at the station.Answer: when we see. The coach has asked when to look at the result. Pick the answer.Pick a move to see the reply.Coach: "See you here at ten."Accept: the next cycle starts tomorrow, and the board will show the result.Coach: "Three months is a long time to wait for one step."Ask again: the steps should be days apart.

Answers the coach can accept: 0 of 5Answers the coach can accept: 1 of 5Answers the coach can accept: 2 of 5Answers the coach can accept: 3 of 5Answers the coach can accept: 4 of 5Answers the coach can accept: 5 of 5

  1. Ask: the target (Coach). The coach asks at the station's board, every morning at the same time, the same five questions in the same order.
  2. Answer: the target (Learner). The coach has asked what the target condition is. Pick the shift lead's answer.
    • Pack each order in 45 seconds, every cycle, with no rework, by the end of the month. Coach: "Good. Now, where are you today?" Accept: a condition at the process, with a number and a date.
    • Get the team faster and better at packing. Coach: "How would we know you got there?" Ask again: a wish, with nothing to measure.
  3. Ask: where now (Coach). Then the coach asks for the facts, and what was learned from the last step.
  4. Answer: where now (Learner). The coach has asked for the actual condition. Pick the answer.
    • Twenty timed cycles yesterday: average 54 seconds, ranging from 47 to 66. Coach: "What made the slow ones slow?" Accept: measured at the station, with the spread as well as the average.
    • Pretty good. The team is working hard. Coach: "Let's go and time a few together." Ask again: an impression, and no number to compare with the target.
  5. Ask: obstacles (Coach). The coach asks what stands between the actual condition and the target, and which one the learner is taking on now.
  6. Answer: obstacles (Learner). The coach has asked about obstacles. Pick the answer.
    • Hunting for the right box size costs about 8 seconds a cycle. That's the one I'm on. Coach: "What's your next step on it?" Accept: one obstacle, seen at the process, with a size.
    • Staffing, the old software, and morale. All of them at once. Coach: "Which one are you working on today?" Ask again: a list too big for one step, and none of it chosen.
  7. Ask: next step (Coach). The coach asks for one experiment and a prediction, so the result can teach something.
  8. Answer: next step (Learner). The coach has asked for the next step and the expected result. Pick the answer.
    • Rack the three common box sizes at arm's reach. I expect about 6 seconds saved. Coach: "When can we see what happens?" Accept: small, testable today, with a prediction to check.
    • Redesign the whole packing area. Coach: "What could you try by tomorrow?" Ask again: a project, and nothing to learn from this week.
    • Try something with the boxes. Coach: "What exactly, and what do you expect?" Ask again: no step named and no prediction.
  9. Ask: when we see (Coach). The last question sets the next meeting at the station.
  10. Answer: when we see (Learner). The coach has asked when to look at the result. Pick the answer.
    • Tomorrow at ten, after twenty timed cycles with the rack. Coach: "See you here at ten." Accept: the next cycle starts tomorrow, and the board will show the result.
    • At the quarterly review. Coach: "Three months is a long time to wait for one step." Ask again: the steps should be days apart.

An example

A distribution center set its packing team a challenge for the quarter: ship every order the day it arrives. The shift lead on packing set a first target condition under it: each order packed in 45 seconds, every cycle, with no rework, by the end of the month. She timed twenty cycles and found an average of 54 seconds, with a spread from 47 to 66.

Her manager coached her for ten minutes each morning at the board beside the station, always with the same five questions. The first obstacle she picked was hunting for the right box. Racking the three common sizes at arm's reach saved 5 seconds, a little less than the 6 she had predicted. The next morning's obstacle was the label printer, which jammed twice an hour. Three weeks and eleven small experiments later, the station ran at 46 seconds, and the shift lead had begun coaching the lead on the next station.

Why it works

Improvement stalls when it depends on projects and workshops. The improvement kata turns it into a daily routine: a target condition far enough out to need learning, the facts of today's condition, one obstacle at a time, and a small experiment with a prediction. The prediction is what makes each step teach something, because a gap between what was expected and what happened is new knowledge about the process.

The coaching kata builds the habit. The same five questions, asked at the process every day, make the learner bring facts, pick one obstacle, and name a next step that can be tested soon. The manager's job shifts from handing out answers to keeping the learner's thinking honest, and over months the questions become the way the learner thinks without being asked.

The thinking behind it

Mike Rother Mike Rother is a researcher of the Toyota production system. In Toyota Kata: Managing People for Improvement, Adaptiveness and Superior Results (McGraw-Hill, 2009) he described what he observed in Toyota's management routines as two practiced patterns: the improvement kata, which works toward a target condition by small experiments, and the coaching kata, a daily routine in which a manager coaches that work. Earlier, with John Shook, he had introduced value stream mapping to Western industry in Learning to See (1999).

How to run it

  1. Set the challenge. Name the direction for the next few months, one that the team cannot yet see how to reach.
  2. Grasp the current condition. Time the process yourself and record the spread as well as the average.
  3. Set the next target condition. Describe how the process should run a few weeks out, with a number and a date.
  4. Pick one obstacle. List what stands in the way, then choose the one to work on now and park the rest.
  5. Run one experiment with a prediction. Write down what you expect before you change anything, so the result can be compared.
  6. Coach at the process every day. Ask the five questions in order: target, actual condition, obstacles, next step, and when we can see the result.
  7. Record each step on a board. Keep the prediction, the result, and what was learned side by side where the work happens.

Where it goes wrong

The questions become a status meeting. Coaches ask them in a conference room and accept impressions. Hold the session at the process, and send the learner to measure when the answer has no number.

The coach supplies the answers. A manager who already knows the fix gives it away, and the learner learns to wait for the answer. Ask the next question instead, even when the step looks weak, and let the experiment show it.

The target is a slogan. Faster and better cannot be tested; without a number and a date, no experiment can be compared with anything. Rewrite the target until a stopwatch could check it.

When to use something else

  • PDCA Cycle. When one change needs a full cycle of plan, test, and review, without a daily coaching routine around it.
  • Kaizen. When the aim is a broad culture of small improvements from everyone, without a single target condition.
  • Value Stream Mapping. When nobody yet knows where in the end-to-end flow the target condition should be set.

Where it sits in SOLVE eX

Go to the source

  • Toyota Kata: Managing People for Improvement, Adaptiveness, and Superior Results, McGraw-Hill, 2009

Back to the catalog