A3 Thinking

Thinking Tools1.2 Writing the Problem StatementAlso known as A3 Problem Solving

From Toyota's production system practice, which credits no single author, as set out by Durward K. Sobek II and Art Smalley (Understanding A3 Thinking, 2008) and by John Shook (Managing to Learn, 2008).

Fits a whole problem onto one A3 sheet (background, current condition, target, root cause, countermeasures, plan, and follow-up) and revises it with peers until the reasoning holds.

Five boxes of an A3 on picking errors, and the coach's question at each

At each of the author's turns, pick what goes in the box: the first-draft version or one that will hold up. The coach's reply shows with a verdict, and the line under the panel counts the boxes that follow from the one before. Illustrative.

Author

Coach

Author

Coach

Select a stage to read it, or play the sequence.The request. Mis-picks at a parts distributor's main warehouse have nearly tripled since spring. The site manager asks a shift lead to write an A3; the operations manager coaches her drafts.Current condition. Pick what the author writes in the current condition box.Pick a move to see the reply.Up from what? Where exactly? Have you watched it happen?A verdict before the facts: nothing for the later boxes to build on.Good. What is different about those five racks?Measured and located: it tells the next boxes where to look.Target condition. Pick what the author writes in the target box.Pick a move to see the reply.How will you know when you are done?No number and no date: nothing can be checked against it.Fine. Now why is it 1.1?A target with a measure: the follow-up box can test it.Go and see. The coach sends the author to watch picking at racks 14 to 18 before she writes the cause.Root cause. Pick what the author writes in the root cause box.Pick a move to see the reply.That is a countermeasure. Why would trained pickers err at only five racks?A solution in the cause box: it does not explain where the errors happen.That explains the racks and the timing. What would stop the swaps?A cause that explains the pattern in the numbers.Countermeasures. Pick what the author writes in the countermeasures box.Pick a move to see the reply.Does the new system stop neighbors being swapped?Expensive, slow, and not aimed at the cause.Both act on the swap. Who owns each, and by when?Countermeasures that act on the root cause.Follow-up. Pick what the author writes in the follow-up box.Pick a move to see the reply.What result would tell you the cause was wrong?A check that cannot fail, so it cannot teach.Good. Bring it to me in four weeks.A check with a trigger: the sheet can be proven wrong.Sign-off. The third draft holds from box to box, and the coach signs it. Within five weeks the error rate is 0.5 percent, and nobody has been retrained.

Boxes that hold up: 0 of 5Boxes that hold up: 1 of 5Boxes that hold up: 2 of 5Boxes that hold up: 3 of 5Boxes that hold up: 4 of 5Boxes that hold up: 5 of 5

  1. The request (Coach). Mis-picks at a parts distributor's main warehouse have nearly tripled since spring. The site manager asks a shift lead to write an A3; the operations manager coaches her drafts.
  2. Current condition (Author). Pick what the author writes in the current condition box.
    • Errors are up and the pickers are getting careless. Up from what? Where exactly? Have you watched it happen? A verdict before the facts: nothing for the later boxes to build on.
    • Mis-picks rose from 0.4 to 1.1 percent of lines since March, mostly in racks 14 to 18. Good. What is different about those five racks? Measured and located: it tells the next boxes where to look.
  3. Target condition (Author). Pick what the author writes in the target box.
    • Reduce errors as much as possible. How will you know when you are done? No number and no date: nothing can be checked against it.
    • Back to 0.4 percent of lines by the end of September, measured weekly from returns. Fine. Now why is it 1.1? A target with a measure: the follow-up box can test it.
  4. Go and see (Coach). The coach sends the author to watch picking at racks 14 to 18 before she writes the cause.
  5. Root cause (Author). Pick what the author writes in the root cause box.
    • The pickers need retraining. That is a countermeasure. Why would trained pickers err at only five racks? A solution in the cause box: it does not explain where the errors happen.
    • A March re-slot put look-alike parts in neighboring bins; most errors are swaps. That explains the racks and the timing. What would stop the swaps? A cause that explains the pattern in the numbers.
  6. Countermeasures (Author). Pick what the author writes in the countermeasures box.
    • Replace the warehouse system next year. Does the new system stop neighbors being swapped? Expensive, slow, and not aimed at the cause.
    • Separate look-alike parts by at least one bin, and add a scan check on forty part numbers. Both act on the swap. Who owns each, and by when? Countermeasures that act on the root cause.
  7. Follow-up (Author). Pick what the author writes in the follow-up box.
    • Review the results in a few months. What result would tell you the cause was wrong? A check that cannot fail, so it cannot teach.
    • Track the weekly rate; if it is not under 0.6 percent in four weeks, revisit the cause. Good. Bring it to me in four weeks. A check with a trigger: the sheet can be proven wrong.
  8. Sign-off (Coach). The third draft holds from box to box, and the coach signs it. Within five weeks the error rate is 0.5 percent, and nobody has been retrained.

An example

Picking errors at a parts distributor's main warehouse had nearly tripled over the spring, and the site manager asked a shift lead to write an A3 on it. Her first draft fit easily: errors were up, pickers were careless, and the countermeasure was retraining.

Her coach, the operations manager, did not edit it. He asked questions: up from what, where exactly, and had she watched it happen. The second draft had numbers: mis-picks had gone from 0.4 to 1.1 percent of order lines since March, and seven in ten were in five rows of racking. Watching those rows for a shift showed why. A March re-slot had put look-alike parts in neighboring bins, and most errors were swaps between neighbors.

The third draft held together from box to box. The target was 0.4 percent by the end of September, measured weekly from returns. The countermeasures were to separate look-alike parts by at least one bin and add a scan check on forty part numbers. Errors were back to 0.5 percent within five weeks, and nobody had been retrained.

Why it works

A problem written on one sheet has nowhere to hide a gap. When the current condition, the target, the root cause, and the countermeasures sit side by side, a reader can see at once whether each box follows from the one before: whether the cause explains the numbers, and whether the countermeasure acts on the cause. A slide deck lets each of those live on its own page, where the break between them goes unnoticed.

The drafts are the method. A coach who asks questions, and leaves the author to answer them on the sheet, makes the author go and look, measure, and reason again. Each draft shows the coach how the author is thinking, which is why the sheet develops people as well as fixing problems. The third draft of a good A3 often blames a different cause than the first, and the countermeasure changes with it.

The thinking behind it

Art Smalley Art Smalley, a former Toyota employee and lean consultant, wrote Understanding A3 Thinking (2008) with the researcher Durward Sobek. Their book is the standard account of Toyota's practice of working a problem through plan, do, check, act on a single sheet of paper, carrying it from the background and current condition through root cause and countermeasures to the follow-up that shows whether the change worked.

John Shook John Shook, a lean management author, described Toyota's A3 process in Managing to Learn (2008). The process puts a problem, its analysis, the countermeasures, and the plan on one large sheet of paper, and he presents it as a way of developing people as well as solving problems, through the exchange between an author and the manager who coaches the drafts. Earlier, with Mike Rother, he wrote Learning to See (1999), the workbook that taught value-stream mapping.

How to run it

  1. Name the problem and why it matters. Write the background in two or three lines, with the cost or the customer it hurts.
  2. Measure the current condition. Give the number, the trend, and where it happens, from data and from watching the work.
  3. Set the target condition. Write the number to reach, by when, and how it will be measured.
  4. Find the root cause. Keep asking why until the cause explains the numbers in the current condition.
  5. Match countermeasures to the cause. List actions that act on the cause, each with an owner.
  6. Write the plan and the check. Give dates, and the result that would send you back to the root cause.
  7. Review each draft with a coach. Let the coach ask questions, and answer them by revising the sheet.

Where it goes wrong

The sheet is written backward from a favored solution. The countermeasure was chosen first, and the root cause box is filled to justify it. Spot it when the cause does not explain the pattern in the current condition; send the author back to watch the work.

The coach edits the sheet. Corrections in red pen fix the document and teach nothing, and the next A3 has the same gaps. Coaches should ask, and authors should write.

The A3 becomes a form. A team fills in every box for every small issue, and the sheet turns into paperwork. Keep it for problems that need joint diagnosis, and let a quick fix stay quick.

When to use something else

  • PDCA Cycle. When the change is small and the team needs a fast loop of predict, try, and check, without a written sheet.
  • DMAIC. When the defect is costly and the causes need formal measurement and statistical analysis before any change.
  • Five Whys. When the problem is narrow and one chain of causes is likely to reach the fix.

Where it sits in SOLVE eX

Go to the source

  • Understanding A3 Thinking: A Critical Component of Toyota's PDCA Management System, 2008
  • Managing to Learn: Using the A3 Management Process, Lean Enterprise Institute, 2008

Back to the catalog