A3 Problem Solving in Manufacturing: Make Change Work

Home » Resources » A3 Problem Solving in Manufacturing: Make Change Work

Estimated reading time: 10 minutes

The corrective action is marked complete. The fixture has been adjusted, the team has been reminded of the procedure, and production is running again. Next week, the same defect returns.

The action happened. The problem was not necessarily solved.

A3 problem solving helps a manufacturing team connect the problem it sees, the evidence for its cause, the change it wants to test and the results that justify keeping that change. The point is not to produce a polished sheet of paper. It is to make a better production decision together.

That decision also needs time on the factory floor. SkyPlanner APS brings an approved production trial into the same schedule as customer orders, reducing the work of fitting improvement activities around separate machine queues. Your team can investigate the problem, agree on a countermeasure and give the experiment a realistic place to happen.

What is A3 problem solving?

A3 problem solving is a structured, collaborative approach to understanding a problem, testing countermeasures and checking their effectiveness. Developed at Toyota, it uses a concise, usually one-page report to make the reasoning visible and support discussion between the problem owner, a coach and the people affected.

The name comes from A3 paper, which measures 297 × 420 mm. The paper size encourages a clear, visual story, but the thinking matters more than the dimensions. A digital page can carry the same reasoning.

Three things belong together:

• A3 thinking: investigate before deciding, and test your explanation against what actually happens.
• The A3 process: involve the right people, work through the evidence and revise your understanding as you learn.
• The A3 report: summarize that reasoning so another person can challenge it and help move the work forward.

A completed report does not prove that the thinking was sound. Nor does an installed countermeasure prove that it worked.

How does A3 relate to PDCA and root cause analysis?

A3 follows the learning logic of Plan–Do–Check–Act (PDCA): understand the situation and plan a change, try it, examine the outcome, then decide what to adopt or investigate next.

Root cause analysis supports the investigation inside that cycle. A fishbone diagram or a 5 Whys discussion can help develop possible explanations, but neither replaces evidence. A3 carries the story further: why this countermeasure, who will test it, how the trial will happen and what result would justify changing normal work.

Use A3 when a recurring or significant problem needs investigation and coordination. It is especially useful when production, quality, engineering and planning see different parts of the same issue. A small, understood problem may need a direct correction rather than a formal A3. An immediate safety or quality risk still needs prompt containment through your established procedures; do not wait for the report.

What should go on an A3 report?

A useful A3 lets someone follow the reasoning without attending every discussion. Layouts and numbers of steps vary. Choose headings that make the following questions easy to answer, rather than treating one template as a universal standard.

Part of the story	What the reader needs to understand
Background and problem	Why this matters, where it occurs and the specific gap to close
Current condition	What happens now, supported by observations and a defined baseline
Target condition	What should change, by when and how success will be measured
Cause analysis	Which explanation the evidence supports, and what remains uncertain
Countermeasure and trial	What change will test that explanation and why it was selected
Implementation	Who owns each action, which resources are needed and when work will happen
Follow-up and learning	What happened against the target, what to retain and what to investigate next

A common arrangement puts the problem and analysis on the left and the proposed change and follow-up on the right. Use a small process sketch, a defect photo or a before-and-after trend where it explains more than a paragraph. Keep supporting records available without forcing every observation onto the page.

The A3 should change as the investigation develops. If a trial disproves the proposed cause, revise the explanation. Do not rewrite the original target to make the result look successful.

Give the story an owner, not just an approval signature

Name one person responsible for moving the A3 forward. That person gathers evidence, involves colleagues and follows up on decisions; they do not have to perform every action or manage every department involved.

A coach or manager helps by testing the reasoning. Useful questions include: What did you observe directly? What else could explain this pattern? Why should this change affect the problem? What would make you reject your explanation?

Discuss the work with operators and other affected people while it is developing. A report presented only after the solution has been chosen invites signatures, not learning. Early dialogue can expose a practical obstacle before it becomes an expensive trial.

A manufacturing example: recurring defects after a changeover

Imagine a machining team repeatedly finding an out-of-tolerance feature in the first parts after changing to one product family. The usual response is to adjust the setup and remind the operator to check it. The team wants an A3 because the problem keeps returning and interrupts work at a shared machine.

Describe the gap without hiding a solution inside it

“Operators need more training” is a proposed explanation and response, not a problem statement. “Machining quality is poor” is too broad to guide an investigation.

A more useful starting point is: “For this product family on this machine, first-off inspections are repeatedly failing after changeover, requiring setup correction before the batch can proceed.” The team then adds the observed frequency, affected period and resulting lost production time from its own records.

Record both the inspection outcome and the exposure: failed first-off checks out of all comparable changeovers, not just a count of defects. Otherwise, a quieter production week could appear to be an improvement simply because fewer changeovers occurred.

Observe the current work and test the explanation

The owner watches changeovers with the operators and quality engineer. The team compares the current setup instruction, the fixture actually used and the point at which each check occurs. It separates observations from hypotheses.

Suppose the investigation suggests that an ambiguous fixture-location instruction permits two different setups. That is still an explanation to test, not an established root cause merely because everyone agrees it sounds plausible.

The team checks whether the proposed difference is present in affected changeovers and absent in comparable successful ones. Any controlled reproduction must follow the plant’s safety and quality requirements. If the observations do not support the explanation, the investigation continues instead of moving straight to retraining.

Agree on a countermeasure and a meaningful test

If the evidence supports the fixture-location explanation, a possible countermeasure is an approved locating aid with an unambiguous setup check. The quality and engineering teams decide whether it is suitable and how to trial it safely.

The A3 states the expected effect before the trial: fewer first-off failures for the defined product family under comparable conditions. It also states the acceptance criteria, observation period and any safeguards, such as maintaining dimensional conformity without creating unacceptable setup time.

Assign an owner to preparing the aid, another to collecting inspection results and a clear decision owner for accepting or rejecting the change. Include the machine time, people and inspection capacity the trial needs. “Test next week” is not yet an executable plan.

Give the approved trial a place in the production schedule

Improvement work often competes with the very orders it is meant to help. In the example, preparing and testing the locating aid needs the shared machining resource while customer jobs still need to ship.

Keeping the trial on a separate action list leaves the planner to reconcile it with the machine queue, the operators’ availability and downstream work. When an urgent order arrives, that reconciliation has to happen again. The trial is easily postponed because its capacity demand is not visible in the production plan.

SkyPlanner makes that coordination easier. Define the approved trial as production work with the required process steps, eligible workstations, durations and dependencies. Arcturus, SkyPlanner’s scheduling engine, places that work alongside customer jobs while considering shifts, workstation capacity and the configured scheduling rules.

The Gantt timeline shows where the trial fits and how the shared workload affects order completion. Instead of correcting separate machine queues by hand, the planner works with one connected schedule and can make the delivery consequences part of the decision.

Your team still decides whether the trial should proceed and which work has priority. Priorities, rules and locks steer the scheduling calculation; putting a trial on the timeline is not a safety or quality authorization.

Once a successful change is approved for regular use, update the relevant production data, such as the agreed setup duration or process route. SkyPlanner then schedules future work using those inputs. This connects the improvement to daily planning instead of leaving the old assumptions in the schedule.

SkyPlanner integrates with any ERP or MES through its open REST API, so this scheduling approach can work alongside the business systems you already use.

Check the effect, not just whether the action was completed

Installing the locating aid closes an implementation task. It does not close the A3.

Compare the trial outcome with the original target using the same definitions as the baseline. In the machining example, examine comparable changeovers, first-off inspection results and setup time. Record changes in product mix or operating conditions that could explain the difference. A faster job on an easier part does not demonstrate that the countermeasure solved the original problem.

SkyPlanner gives that review useful production context. Job Report shows the actual time logged against a job, who worked on it and the workstation used. All Timelogs provides timestamped events, while Order Item Report brings together duration, logged time and their difference at the order-item level.

These records help the team connect observations to the work that actually ran, rather than comparing unidentified totals from different weeks. Use them alongside the inspection results and direct process observations needed for the A3. Logged time alone cannot establish why a defect occurred or prove that the countermeasure caused an improvement.

Preserve the original baseline and trial records before changing future planning assumptions. A revised estimate is useful for the next schedule, but it must not replace the reference against which the experiment is judged.

If the trial does not meet the target, return to the evidence. Was the proposed cause wrong, was the countermeasure applied differently than intended, or did another condition change? Record what you learned and plan the next test. Do not close the report simply because the action list is empty.

Make the accepted change part of normal work

When the evidence supports the change and the responsible team accepts it, update the relevant work instructions, setup checks and training. Confirm that affected shifts can use the new method, and set a follow-up point to check that the improvement holds. Share the learning with similar processes where it may apply, rather than assuming the same answer fits every machine.

The scheduling handoff matters too. A new method cannot help reliably if production keeps following an outdated queue.

At electrical equipment manufacturer UTU, department-specific spreadsheets once made it difficult to communicate production-sequence changes to the shop floor. Its SkyPlanner experience describes a shared planning view across the factory and updated work queues reaching the line in seconds. That is useful evidence for getting a changed plan into daily work, rather than leaving people to coordinate separate lists.

Start with one problem that keeps coming back

Choose a recurring production gap with a clear boundary and a named owner. Observe the work with the people doing it, establish a comparable baseline and use the A3 to make the proposed explanation open to challenge. Agree on the trial and the evidence that will decide whether the change deserves to stay.

Use SkyPlanner to turn that agreed production work into a realistic schedule alongside customer orders. You save the effort of repeatedly matching an improvement list to separate machine queues, and the team can see what the trial means for production before committing to it.

Book a demo with one approved trial, the shared resource it needs and the customer jobs competing for that resource. See how SkyPlanner brings them into a connected plan. Or start a trial with demo data or the AI-assisted Getting Started in SkyPlanner Assistant to explore the scheduling workflow yourself.