Coaching kata, A3, after-action review: the questions that run a strategy
A strategy is not run by its plan but by the conversations held around it. Four routines structure those conversations with questions rather than instructions: the coaching kata (Rother, 2009) for the owner of an initiative, the A3 dialogue (Shook, 2008) for a problem, the after-action review (US Army, 1993) at the end of a period, and catchball (Akao, 1991) between levels before objectives are set.
They share one principle, older than all of them: learning comes from stating what you expect, acting, and studying the difference. This page goes back to that principle, says where each routine comes from and what it is for, and says what an AI agent may take on: asking the questions and bringing the facts, never the answers.
The principle under all four: expect, act, study the difference
In 1939 Walter Shewhart described production as a three-step scientific process: specification, production, inspection. W. Edwards Deming turned it into a cycle and taught it in Japan from 1950; Japanese practitioners made it Plan-Do-Check-Act. Deming himself preferred Plan-Do-Study-Act and asked that it be called that: in English, to check is also to hold back, and the third step is about learning, not control. Moen and Norman trace this history in detail.
Every routine on this page is that cycle applied to a conversation. Each asks what was expected before asking what happened, because a result with no prior expectation teaches nothing: there is no difference to study.
Hoshin Kanri applies the same cycle to the policy of the year, as the Hoshin Kanri process shows. The routines below are how the cycle is kept alive between two reviews, and inside them.
The coaching kata: five questions to the owner of an initiative
Mike Rother studied from 2004 to 2009 how Toyota manages people for improvement, and described two routines: an improvement kata (understand the direction, grasp the current condition, set the next target condition, experiment toward it) and a coaching kata, a short dialogue a manager repeats with the person doing the improving.
Worded for a strategy initiative, the five questions are: what are we trying to reach; where are we now; what is in the way, and which obstacle are you working on now; what is your next step, and what do you expect from it; when can we see what we learned from that step. The wording here is ours; the pattern is Rother’s, and his practice guide carries his own card.
Two things matter more than the wording. One obstacle at a time, and a small step whose expected result is stated before it is taken. And the coach does not give the solution: the aim is that the owner learns to find the way, and keeps the initiative.
A precision that is often lost: “kata” is Rother’s word for what he observed. It is not an official Toyota procedure, and he presents it as a set of practice routines for other organizations. Spear and Bowen had described ten years earlier, independently, that every change at Toyota is run as an experiment under the guidance of a teacher.
The A3 dialogue: a problem on one page, discussed with a mentor
An A3 holds a problem on a single sheet: background, current condition, goal, analysis, countermeasures, plan, follow-up. John Shook’s Managing to Learn shows that the sheet is the lesser half. The other half is the dialogue between its author and a mentor who asks, sends the author back to the facts, and does not rewrite the sheet.
Use it when a result is off and nobody yet knows why. A coaching cycle moves an initiative forward step by step; an A3 stops to understand one problem in depth. The prompts and the limits of an assistant on that work are in A3 problem solving with AI.
The after-action review: four questions at the end of a period
The US Army formalized the after-action review for training exercises and published a leader’s guide to it in 1993 (Training Circular 25-20). It is a professional discussion in which the participants discover for themselves what happened and why. Morrison and Meliza wrote its history and foundations in 1999.
Worded for a strategy review: what did we expect; what happened; why the difference; what do we keep, and what do we change. The first question is answered from the decisions of the previous review, the second from the readings. The third is about the work, never about a person. What comes out of the fourth is a decision, and is recorded as one.
It takes ten minutes at the end of a monthly review and it is what turns a review from a reading of figures into a lesson. How the rest of the review is prepared is in preparing a strategy review.
Catchball: questions between levels, before the objectives are set
Catchball belongs to Hoshin Kanri itself (Akao, 1991). Leadership proposes objectives, the level below answers with what it can reach and what it needs, and the proposal goes back up amended, until the targets are ones the people who must reach them have discussed. It is a dialogue across levels, where the three routines above are dialogues inside one.
Its question is not “do you agree” but “what would it take”. The full mechanics, with a worked example, are in catchball in Hoshin Kanri.
- Before the year: catchball, to set objectives people have discussed.
- Between two reviews: the coaching kata, initiative by initiative.
- When a result is off and the cause unknown: an A3 and its dialogue.
- At the end of each period: the after-action review, inside the review.
What makes the questions work: going to see, and safety to answer
A question asked far from the facts gets an opinion. Toyota’s own account of the Toyota Way 2001 places genchi genbutsu, going to the source to see, among its five concepts, under two pillars: continuous improvement and respect for people. Asking instead of telling is the practical form of that respect.
And a question gets a true answer only where it is safe to give one. Amy Edmondson studied 51 work teams in a manufacturing company and found that teams whose members felt safe to take interpersonal risks showed more learning behaviour, which in turn went with performance. A coaching routine used to catch people out destroys what it needs.
These routines have a common ancestor in the Training Within Industry service of the United States, which from 1940 to 1945 taught supervisors three skills as short repeatable programmes: instructing a job, improving its method, and handling relations with people.
What an AI agent may do with them
An agent connected to the matrix is well placed to ask these questions, for one reason: it has the facts in front of it. Before each question it can say what the strategy holds: the expected result of the initiative, the last readings of its indicator and their trend, the blockers noted last time, the decisions of the previous review.
And it is badly placed to answer them. The method is that the owner finds the next step and leadership decides; an agent that proposes the step takes the initiative away from its owner. On hoshin.app the guidance given to agents says so in those terms: if the person asks what to do, give the question back, or offer a fact.
Concretely, the Hoshin Kanri plugin for Claude carries a skill that runs a coaching cycle on an initiative, and its review skill offers the four after-action questions before a review is recorded. With the person’s agreement the agent writes the cycle down in their words: the obstacle chosen, the step, what is expected, the date to look at the result. What the app does not hold yet: a target condition for the coming weeks and the record of each experiment have no field of their own; they live in the notes of the initiative.
Frequently asked questions
What are the five coaching kata questions?
In Mike Rother’s coaching kata, a coach asks about the target condition, the actual condition now, the obstacles and the one being addressed, the next step and what is expected from it, and when the result can be seen. The exact card is in his Toyota Kata Practice Guide (2017). The questions on this page are reworded for a strategy initiative.
Is Toyota Kata an official Toyota method?
No. “Kata” is the word Mike Rother chose for the routines he described after studying Toyota from 2004 to 2009. Toyota’s own published account of its principles is the Toyota Way 2001: continuous improvement and respect for people, with genchi genbutsu among its concepts.
What is the difference between a coaching kata and an after-action review?
The coaching kata looks forward, with one person, on one initiative: the next step and what is expected from it. The after-action review looks back, with a team, on a period: what was expected, what happened, why. Both rest on the same cycle of stating an expectation and studying the difference.
PDCA or PDSA?
Both name the same cycle. Plan-Do-Check-Act is the form that took hold in Japan in the 1950s; Deming preferred Plan-Do-Study-Act, because the third step is about learning from the result rather than controlling it. Moen and Norman (2010) trace both back to Shewhart’s 1939 description.
Can an AI agent coach a team?
It can hold the routine: ask the questions in order, bring the facts of the strategy before each one, and keep a written trace of what was said. It should not answer in the person’s place, choose the next step or judge the person. The value of the routine is that the owner learns the way.
Sources
- Ronald Moen and Clifford Norman, “Circling Back: Clearing up myths about the Deming cycle”, Quality Progress, November 2010
- Mike Rother, Toyota Kata, McGraw-Hill, 2009; on the research of 2004 to 2009: “A Kata Classic”, Lean Enterprise Institute, 2016
- Mike Rother, The Toyota Kata Practice Guide, McGraw-Hill, 2017
- Steven Spear and H. Kent Bowen, “Decoding the DNA of the Toyota Production System”, Harvard Business Review, September-October 1999
- John Shook, Managing to Learn, Lean Enterprise Institute, 2008
- John E. Morrison and Larry L. Meliza, Foundations of the After Action Review Process, US Army Research Institute, 1999
- Yoji Akao (ed.), Hoshin Kanri: Policy Deployment for Successful TQM, Productivity Press, 1991
- Toyota Motor Corporation, 75 Years of Toyota: the Toyota Way 2001
- Amy Edmondson, “Psychological Safety and Learning Behavior in Work Teams”, Administrative Science Quarterly, 44(2), 1999
- US War Manpower Commission, More Production Through Training: The Training Within Industry Program