A3 Problem Solving

A3 is a disciplined problem-solving and coaching process expressed as a concise visual story. Its value is not fitting text on one sheet; it is making the reasoning from purpose to evidence, cause, action and learning easy to challenge.

In one minute

An A3 usually connects:

1. Background: why the issue matters and for whom.
2. Current condition: what is happening, where, when and at what scale.
3. Target condition: a measurable outcome within a defined boundary and date.
4. Analysis: testable causal mechanisms, not labels or blame.
5. Countermeasures: actions linked to causes, with safeguards.
6. Plan: owners, timing, dependencies and decision rights.
7. Follow-up: outcome, process, side-effect and learning measures.

The sequence is iterative. New evidence may change the current-condition picture, causal model or target.

Best for: a bounded performance gap that needs shared reasoning and owned experiments.
Avoid when: immediate harm needs containment first, the issue is too broad to bound, or leadership has already imposed an untestable solution.

The problem it addresses

Teams often jump from a symptom to a favourite fix. Status decks separate data, analysis and decisions across many files, while lessons disappear after implementation. A3 creates one navigable evidence story and a coaching dialogue around it.

The sheet is not evidence that the problem was solved. Improvement claims require a baseline, comparison window, process evidence, side-effect monitoring and an explanation of alternative causes.

When to use it

Use A3 Problem Solving when:

• a recurring operational gap has a measurable current condition;
• several functions own different parts of the causal system;
• a proposed countermeasure needs an explicit reasoning chain;
• a team needs coaching in problem-solving rather than a completed answer;
• improvement should be tested through PDCA/PDSA;
• learning and standardisation must survive beyond the project.

When not to use it

Do not use A3:

• instead of containing an active safety, legal or financial incident;
• to compress a complex portfolio into one page;
• as a form completed after the solution is chosen;
• to label a person or team as the root cause;
• when measurement would expose personal data without authority;
• to claim attribution from a before/after comparison alone.

Inputs required

• a bounded gap, affected users and accountable owner;
• baseline measures with definitions and denominators;
• direct observation of the work and exception paths;
• process, demand, timing and qualitative evidence;
• causal hypotheses and disconfirming tests;
• constraints, safeguards and decision rights;
• capacity for follow-up after implementation.

Step-by-step process

1. Clarify the background

State the purpose, affected groups, strategic or operational relevance and why action is timely. Keep solution language out.

2. Observe the current condition

Go to the work where practical. Show the process, variation, frequency and impact. Define each metric and separate facts from interpretations.

3. Define the target condition

Specify the desired outcome, boundary, measure, date and guardrails. The target should close a meaningful gap without shifting harm elsewhere.

4. Analyse causal mechanisms

Use Root Cause Analysis, Five Whys, Ishikawa or another appropriate method. For each causal claim, record supporting and disconfirming evidence.

5. Select countermeasures

Link each proposed action to a causal mechanism. Compare effectiveness, feasibility, reversibility, side effects and required authority.

6. Plan implementation

Name owners, dates, dependencies, training, communication, escalation and stopping rules. Separate a countermeasure test from full rollout.

7. Define follow-up

Measure outcome, process adherence, balancing effects and differentiated effects on affected groups. Set the comparison window before acting.

8. Test and learn

Run the bounded change, review evidence and update the A3. Adopt, adapt or abandon; do not rewrite the original hypothesis after the result.

9. Standardise carefully

If evidence supports the change, update standard work, controls and training. Preserve exceptions, residual risk and a review trigger.

AI automation lens

AI can summarise authorised observations, chart defined measures, connect countermeasures to recorded causes and flag missing owners or follow-up signals.

It must not:

• invent observations or causal evidence;
• mine employee data to assign blame;
• optimise a local metric while hiding downstream harm;
• turn correlation into a root-cause claim;
• rewrite the hypothesis after results are known;
• approve a safety-critical countermeasure.

Visual model

Text alternative: one evidence story connects purpose and current condition to a target, tested causes, linked countermeasures, an owned plan and follow-up that updates the understanding.

Interactive example

Scenario

A shared-service team reports that 14% of supplier invoices are returned for correction. Leaders propose retraining everyone. Observation shows most errors come from one new intake path where purchase-order fields are mapped ambiguously; manual rework hides the pattern.

Worked answer

Contain payment risk and define the baseline by channel. Set a target for valid first-pass invoices while monitoring late payment and supplier burden. Test the mapping and validation mechanism before system-wide training. Compare the new path with unaffected paths and retain the original hypothesis and results.

Facilitation notes

• The A3 owner does the thinking; the coach asks for evidence and coherence.
• Use diagrams and charts where they reveal the condition better than prose.
• Invite people who perform and receive the work.
• Keep countermeasures out of the current-condition section.
• Review left-to-right logic and right-to-left traceability.

Expected output

• a bounded problem and evidence-based current condition;
• a measurable target with safeguards;
• causal hypotheses with tests;
• countermeasures traceable to mechanisms;
• an owned implementation and escalation plan;
• outcome, process and balancing measures;
• a versioned learning record and standardisation decision.

Common mistakes

1. One-page status report. A3 is reasoning and coaching, not compression.
2. Solution in the problem statement. This narrows investigation prematurely.
3. Root cause as a person. Human error is a starting observation, not a mechanism.
4. Action list without causality. Every countermeasure needs a reason it should affect the gap.
5. No balancing measure. Local improvement can shift cost or harm.
6. Retrospective certainty. Preserve what was believed before the test.

Quality checklist

• The gap, boundary, affected groups and owner are explicit.
• Current-condition metrics have definitions and denominators.
• Direct observation and exception paths are represented.
• The target includes a date and harm guardrails.
• Causal claims have confirming and disconfirming evidence.
• Countermeasures trace to causal mechanisms.
• Outcome, process and balancing measures are defined before action.
• Learning, residual risk and the next review are recorded.

Template

Knowledge check

Question: What most clearly distinguishes an A3 countermeasure from an action item?

Answer: It is linked to a testable causal mechanism and evaluated with outcome, process and balancing evidence.

Related tools

• Root Cause Analysis
• Five Whys
• Ishikawa Diagram
• PDCA/PDSA Cycle

References

1. Lean Enterprise Institute. “A3 Report.” Lean Lexicon (opens in a new tab). Accessed 22 September 2026.
2. Shook, John. Managing to Learn: Using the A3 Management Process to Solve Problems, Gain Agreement, Mentor, and Lead. Lean Enterprise Institute, 2008.
3. Sobek, Durward K. II, and Art Smalley. Understanding A3 Thinking. Productivity Press, 2008.
4. Poksinska, Bozena, et al. “The Daily Work of Lean Leaders—Lessons from Manufacturing and Healthcare.” Total Quality Management & Business Excellence, 2020. DOI (opens in a new tab). The literature notes limited systematic research on A3 practice.
5. Dunsford, Joshua, and Erica Reimer. “Relationship-Centred Care and A3 Improvement in Clinical Practice.” PLOS ONE, 2013. Article (opens in a new tab). Application evidence does not establish universal effects.

Method profile

• Primary output: a concise, versioned evidence story and owned learning cycle.
• Decision level: bounded operational or cross-functional improvement.
• Evidence strength: established Lean practice with case applications; systematic comparative evidence for the A3 process itself is limited.
• Review trigger: new baseline evidence, failed causal test, side effect, changed process or completed follow-up window.