Worked examples · 3 complete A3 reports
A3 examples: three worked A3 reports, with a coach's notes
An A3 example is a filled-in A3 report: one problem, proposal or project told on a single large sheet, the left side describing the situation and the right side saying what to do and how to check it. Below are three complete A3s from plant settings, one of each kind (problem solving, proposal and status), each with a coach's note on every box saying why it works and the weak version reviewers see most often.
Illustrative examples. All three are illustrative: written from the method with realistic numbers, not taken from a real company. Every number on them adds up, and you can redo each calculation from the figures shown. Want the blank sheet? The A3 problem solving template is free in PDF and Excel.
Three kinds of A3, three storylines
Same sheet, different job. Pick the kind before you write: a problem-solving A3 that is really a request for money reads badly, and so does a proposal with no analysis behind it.
Problem-solving A3
- Use it for:
- A gap between the standard and what happens: scrap, defects, downtime, injuries.
- Left side:
- Background, current condition with data, goal, root cause confirmed with evidence.
- Right side:
- Countermeasures tied to causes, plan with owners and dates, follow-up against the goal.
- Ends with:
- A checked result. If the goal is missed, a next step and the A3 stays open.
Proposal A3
- Use it for:
- A change that needs someone's approval: money, a layout, a new way of working.
- Left side:
- Background, current condition, goal, analysis of what drives the current situation.
- Right side:
- Options compared on the same arithmetic, a recommendation, plan, unresolved issues, how it will be checked.
- Ends with:
- A decision: approved, approved in part, or sent back with questions.
Status A3
- Use it for:
- A project or a KPI recovery that has been running for a while, reported to a review.
- Left side:
- Background, current status with the trend, target and gap, what was done in each PDCA cycle.
- Right side:
- Remaining issues, next steps, a forecast, and any help needed.
- Ends with:
- Agreement on the next cycle, and help where the owner cannot act alone.
A3 1 of 3 · Problem-solving A3
Problem-solving A3: leak test failures on a pump line
It answers: Why does this gap exist, and what will close it? Look at how the evidence narrows: a Pareto of teardown findings, a split by hour of the shift, six candidate causes checked and five ruled out, then a 5 Whys that ends at a system, not a person. The result misses the goal, and the sheet says so.
Theme
Leak test failures on pump line 3
- Owner
- Quality engineer, line 3
- Coach
- Plant quality manager
- Team
- Line 3 team leader, two assemblers (one per shift), maintenance technician
- Revision
- Rev 4, end of week 10
Plan: the problem and its cause
1Background
Line 3 builds a small centrifugal pump: a cast aluminium housing and a cover sealed with an O-ring. Every pump is air leak tested at the end of the line. A failed pump is torn down, rebuilt with a new O-ring and tested again.
Leak failures are the biggest rework item on the line. A rebuild takes about 12 minutes, so the 432 failures in weeks 1 to 4 cost 86.4 hours of rework (432 × 12 min = 5,184 min). The test checks every pump, so this is a cost and capacity problem, not a customer escape.
2Current condition
Weeks 1 to 4: 9,600 pumps tested, 432 failed, a failure rate of 4.5%. The quality plan allows 1.0%.
- O-ring pinched or rolled324 (75.0%)
- Porosity in the casting54 (12.5%)
- Cover screws loose32 (7.4%)
- No fault found on retest22 (5.1%)
- Hours 1 to 481 (1.7%)
- Hours 5 to 8243 (5.1%)
Same pattern on both shifts and for all six assemblers.
At the station in week 5 we watched 20 assemblies in hour 2 and 20 in hour 7. In hour 7 the lubricant film on the O-ring was thin or patchy on 11 of 20; in hour 2, on 1 of 20.
3Goal
Leak test failures at or below 1.0% of pumps tested, every week for four weeks in a row, by the end of week 12. Measured from the leak tester log: failures divided by pumps tested.
4Root cause analysis
| Candidate (6M) | How we checked | Result |
|---|---|---|
| Material: O-ring lot | How we checked: Ran a different O-ring lot for two days (960 pumps) | Result: Same pattern. Ruled out |
| Material: groove depth | How we checked: Measured 30 housings from three casting lots | Result: All in tolerance. Ruled out |
| Measurement: leak tester | How we checked: Leak and no-leak master parts, three times a shift for a week | Result: Right every time. Ruled out |
| Man: one assembler or shift | How we checked: Split failures by assembler and shift | Result: Spread across all six and both shifts. Ruled out |
| Mother nature: afternoon heat | How we checked: Second shift runs into the cooler evening | Result: Same rise late in that shift. Ruled out |
| Machine: lubricant dose | How we checked: Weighed one shot every hour for two shifts | Result: 0.20 g in hour 1, 0.07 g by hour 7; minimum is 0.15 g. Confirmed |
- Why do pumps fail the leak test? The O-ring is pinched between cover and housing. Evidence: Teardown: 324 of 432.
- Why is it pinched? With too little lubricant the O-ring rolls out of the groove as the cover is pressed on. Evidence: Trial: 50 covers fitted with dry O-rings, 9 pinched; 50 with a full shot, none.
- Why too little lubricant? The bench dispenser is gravity fed, and the shot shrinks as the cartridge empties. Evidence: Shot weight 0.20 g full, 0.07 g by hour 7.
- Why does the cartridge run low? The work instruction says to fill it at the start of the shift only. Evidence: Work instruction, step 4.
- Why does it say that? It was written for the old 2 kg pail, which lasted a week. When the bench dispenser replaced the pail last year, the instruction was not reviewed: the equipment change checklist has no step for it. Evidence: Change record for the dispenser swap; checklist reviewed.
Root cause: A process change (the new dispenser) went in without updating the standard work, because the change checklist does not ask for it.
Do, check, act: what we did and what happened
5Countermeasures
| Cause | Countermeasure | Type | Expected effect |
|---|---|---|---|
| Dose falls late in the shift | Countermeasure: Team leader refills the cartridge at mid-shift (week 5) | Type: Containment | Expected effect: Stops the late-shift rise while the fix is made |
| Refill only at shift start | Countermeasure: Refill at shift start and after 120 pumps; minimum line marked on the cartridge; in the station standard work and the team leader's hourly check | Type: Root cause | Expected effect: Dose stays at or above 0.15 g all shift |
| Change checklist has no standard work step | Countermeasure: Add "review work instructions and check sheets" to the line's equipment change checklist, signed by the area engineer | Type: Root cause | Expected effect: The next equipment change updates the standard |
| Thin dose found only at the leak test | Countermeasure: Shot weight on the station check sheet every hour for four weeks, then weekly if all in range | Type: Detection | Expected effect: A thin dose is seen before it builds a leak |
Out of scope here: Porosity (54) goes to the casting supplier through a supplier corrective action request. Loose cover screws (32) go to the line's next kaizen.
6Implementation plan
| Action | Owner | Due | Status |
|---|---|---|---|
| Mid-shift refill (containment) | Owner: Team leader | Due: Week 5 | Status: Done |
| Mark minimum line on cartridges | Owner: Maintenance technician | Due: Week 6 | Status: Done |
| Update work instruction and station standard work; train both shifts | Owner: Quality engineer | Due: Week 6 | Status: Done |
| Hourly shot weight on the station check sheet | Owner: Team leader | Due: Week 6 | Status: Done |
| Standard work review step in the equipment change checklist | Owner: Area engineer | Due: Week 8 | Status: Done |
| Supplier corrective action request on porosity | Owner: Supplier quality engineer | Due: Week 8 | Status: Sent; reply due week 11 |
| Share the refill rule with lines 1 and 4 (same dispenser) | Owner: Quality engineer | Due: Week 11 | Status: Open |
7Follow-up
Prediction, written in week 6: Pinched O-ring failures close to zero. The other causes run at about 27 a week (porosity, screws, no fault found), so the line should settle at about 1.1% (27 of 2,400): better, but above the 1.0% goal until porosity is fixed.
| W1 | 4.4% |
|---|---|
| W2 | 4.7% |
| W3 | 4.5% |
| W4 | 4.5% |
| W5 | 2.1% |
| W6 | 1.5% |
| W7 | 1.3% |
| W8 | 1.2% |
| W9 | 1.2% |
| W10 | 1.3% |
Result, weeks 7 to 10: 120 failures in 9,600 pumps, 1.25%. Pinched O-rings fell from 324 to 12 (96% fewer). Close to the prediction, so the cause was understood. The goal is not met: porosity is now the top cause (54 of 120, 45%).
- Standardised: refill rule in the station standard work; standard work review in the change checklist.
- Next: porosity through the supplier request; this A3 stays open until four weeks in a row at or below 1.0%.
Coach's notes: why each box works
Why it works: Two short paragraphs answer why anyone should care: what the product is, what a failure costs in hours, and that no customer is at risk. The arithmetic is shown, so a reviewer can check it.
The weak version: Background that tells the history of the line or restates the problem. If a reader cannot say why this problem matters more than the next one, the box has not done its job.
Why it works: Counted facts, a Pareto that picks the biggest cause, and a second cut (time into the shift) that turns a vague problem into a pattern. The last line is direct observation at the station, which is where Sobek and Jimmerson say the current condition has to come from.
The weak version: "Leak failures are high and getting worse" with no count, no period and no split. Without a baseline there is nothing to set a goal against or to check the result against.
Why it works: One measure, a number, a date, and where the data comes from. Anyone can tell in week 12 whether it was met.
The weak version: "Reduce leaks" or "improve first pass yield". Not measurable, so nobody can say when the A3 is finished.
Why it works: Every candidate cause has a check and a result, five are ruled out with data, and each why carries its evidence. The chain stops at a system the team can change (the change checklist), not at a person. The dry-versus-lubricated trial confirms the cause before money is spent.
The weak version: A 5 Whys that stops at "operator did not apply enough lubricant" and a countermeasure of "retrain operators". The data shows all six assemblers and both shifts, so the cause is the method, not the people.
Why it works: Each row starts from a confirmed cause, says whether it contains, fixes or detects, and predicts its effect. Containment is labelled as containment, so nobody mistakes the mid-shift refill for the fix. The causes this A3 will not touch are named, with where they went.
The weak version: A list of good ideas (new dispenser, poka-yoke, retraining, extra inspection) with no link to the causes found. Countermeasures that do not answer a confirmed cause are guesses, however sensible.
Why it works: Every action has one owner (a role, not "team"), a week and a status, so the sheet doubles as the progress check at each review.
The weak version: Owner "all" or "quality", dates "ASAP" or "ongoing". If nobody in particular owns an action, nobody does it.
Why it works: A prediction written before the results, then the actual result next to it. Because they match, the team knows its understanding was right; because the goal is still missed, the A3 stays open with a named next step instead of being declared a success.
The weak version: A follow-up box that says "complete" with ticks against the actions. Finishing the actions is not the same as solving the problem; only the measure against the goal says that.
A3 2 of 3 · Proposal A3
Proposal A3: a changeover kaizen on a filler
It answers: What change do we want approved, and is it worth it? Look at the options table: three options on the same arithmetic, savings counted only where money is really saved (overtime that stops), and the recommendation to buy the cheap option first and decide on the expensive one with measured data.
Theme
Changeovers on filler line 2 take 95 minutes and push work into Saturday overtime
- Owner
- CI lead
- Decision by
- Plant manager
- Consulted
- Line 2 lead and crew, maintenance planner, production planner, quality technician
- Revision
- Rev 3, ready for decision
Why change: the situation and what drives it
1Background
Line 2 fills a liquid product into 500 ml and 1 litre bottles on a rotary filler. Customers order both sizes every week, so the plan runs 6 changeovers a week. Demand has grown and the line now runs Saturday overtime to keep up: 6 hours a week on average over the last 48 working weeks (288 hours).
The production planner expects the same order pattern next year, and cutting overtime is one of the plant's goals this year.
2Current condition
Changeover is timed from the last good bottle of the old size to the first good bottle of the new size at full speed. Twelve changeovers timed in weeks 1 and 2 averaged 95 minutes (78 to 121). At 6 a week that is 570 minutes, 9.5 hours of a stopped line every week.
- Fetch change parts, tools and next bottles20 min
- Remove and fit bolted change parts35 min
- Adjust and trial run until fill and caps are right25 min
- Flush and clean the product path15 min
In the videos the crew walked to the parts store twice per changeover, and the guide rails were set by eye, so the trial run took three to six attempts.
3Goal
Average changeover at or below 65 minutes by week 8, measured on every changeover; and Saturday overtime down by at least 3 hours a week, averaged over weeks 5 to 12.
4Analysis
- Fetching (20 min) is done while the line is stopped but could be done while the old run finishes. This is the separation of internal and external work in Shingo's SMED method.
- Adjusting (25 min) is long because nothing marks the right rail positions for each size, so the crew sets them by eye and trials until it works.
- Bolting (35 min) is internal work that only better fastening or a second set of parts can shorten.
- Flushing (15 min) is set by the product quality procedure and stays as it is in this proposal.
What we propose, what it costs and how we will check
5Options and recommendation
A: SMED workshop. Checklist, changeover cart with shadow board, parts and bottles staged while the old run finishes, rail settings marked
- One-off cost
- $5,000
- Changeover after
- 65 min
- Saved per changeover
- 30 min
- Hours a year
- 144
- Overtime avoided a year
- $25,920
- Payback
- 2.3 months
B: A plus a second set of change parts. The next set is assembled and preset off the line before the stop
- One-off cost
- $45,000
- Changeover after
- 55 min
- Saved per changeover
- 40 min
- Hours a year
- 192
- Overtime avoided a year
- $34,560
- Payback
- 15.6 months
C: A plus quick-release fasteners. The filler maker's retrofit kit replaces bolts on the change parts
- One-off cost
- $21,500
- Changeover after
- 50 min
- Saved per changeover
- 45 min
- Hours a year
- 216
- Overtime avoided a year
- $38,880
- Payback
- 6.6 months
How the numbers work: hours a year = minutes saved × 6 changeovers × 48 weeks ÷ 60. Each hour is worth the overtime it removes: 4 crew × $45 an hour = $180. Counted only up to the 288 overtime hours worked now; hours beyond that are capacity, worth money only when demand needs them. Option A: 30 × 6 × 48 ÷ 60 = 144 hours × $180 = $25,920; $5,000 ÷ $25,920 × 12 = 2.3 months.
Also looked at: a new filler with automatic size change. Not costed here; it needs a capital request and stays on the list for when this filler is replaced.
Recommendation: Approve option A now ($5,000). It gets two thirds of option C's saving (30 of 45 minutes) for under a quarter of the cost. Decide on C in week 8 with measured data: on its own, the extra 15 minutes from quick-release fasteners cost $16,500 and pay back in 15.3 months. Option B pays back slowest (15.6 months) and is not recommended.
6Plan
| When | Action | Owner |
|---|---|---|
| Week 1 | Action: Video two more changeovers; split internal and external steps with the crew | Owner: CI lead |
| Week 2 | Action: Build the changeover cart and shadow board | Owner: Maintenance planner |
| Week 3 | Action: Three-day workshop: checklist, staging during the run, marked rail settings | Owner: CI lead with both shifts' crews and the quality technician |
| Week 4 | Action: Run changeovers to the new standard and log every one | Owner: Line 2 lead |
| Week 4 | Action: Cut Saturday hours to match the time saved | Owner: Production planner |
| Week 8 | Action: Review results; decide on option C | Owner: Plant manager and CI lead |
7Unresolved issues
- The saving is banked only if the planner removes Saturday hours. Agreed in principle; reviewed in week 4.
- Flush time is set by the quality procedure. The quality technician will look at whether a shorter, validated flush is possible, as a separate request.
- First-bottle fill and cap checks stay as they are.
8Follow-up
- Every changeover logged with start and stop by the definition above; weekly average against 65 minutes.
- Saturday overtime hours per week, from the payroll report.
- Prediction: 60 to 70 minutes average from week 5. Above 70, watch the video again before buying anything.
Coach's notes: why each box works
Why it works: It says why now (demand grew, overtime is a plant goal) and gives the one number the cost case will rest on: overtime hours actually worked.
The weak version: Opening with the solution ("We need quick-release tooling on line 2"). Sobek and Jimmerson's rule for the theme applies to the whole left side: describe the problem, not the answer you already like.
Why it works: The measure is defined before it is quoted (last good bottle to first good bottle), the average comes with its range, and the 95 minutes are broken into elements from video, which is what the options will act on.
The weak version: "Changeovers take about an hour and a half." An estimate from memory hides the spread and gives the options nothing to work on.
Why it works: Two measures, because the second one is where the money is: a faster changeover saves nothing if Saturday shifts still run.
The weak version: A goal stated only as the improvement ("SMED on line 2"). That is an activity; the goal is what changes in the numbers.
Why it works: Each element gets a cause, and the analysis says which element no option will touch and why. That keeps the promise in the goal honest.
The weak version: Skipping analysis because "everyone knows changeovers are slow". In a proposal A3 the analysis is what lets the reader judge whether the options can deliver the goal.
Why it works: Options are compared on one formula, the formula is shown, and savings are counted only where cash really changes (overtime that stops). The incremental view of C over A is the line that makes the recommendation sound: the cheap step first, the expensive one decided with data. This is the set-based thinking Shook gives a chapter to.
The weak version: One option presented as the answer, or savings that count every freed hour at full labour cost when no one's pay changes. Finance will find it, and the next A3 starts with less trust.
Why it works: Short, dated and owned, and it includes the step that turns time saved into money saved (the planner cutting Saturday hours).
The weak version: A plan that ends at "workshop held". The saving only exists once the overtime plan changes, so that step belongs on the sheet.
Why it works: It names what could stop the saving and who is dealing with it. A decision-maker can approve knowing the risks.
The weak version: Leaving the box out to make the proposal look clean. The issues come up in the meeting anyway, and then the author looks unprepared.
Why it works: The check is set up before approval, with a prediction and a rule for what to do if it misses. The decision on option C will rest on these numbers.
The weak version: "We will monitor the results." Without a measure, a source and a prediction, the week 8 review becomes a matter of opinion.
A3 3 of 3 · Status A3
Status A3: on-time shipping off target
It answers: Where does this stand, what have we learned and what happens next? Look at the PDCA table: each cycle has its own plan, do, check and act, and the check is a number. The trend shows the gains and the plateau honestly, and the forecast shows exactly how much has to go to hit the target.
Theme
On-time shipping from the finished goods pick area: 97.3% in September against 98%
- Owner
- Warehouse manager
- Audience
- Monthly operations review
- Opened
- April, after two months below target
- Status as of
- End of September
Where we stand and what we have done
1Background
Finished goods are picked, packed and handed to carriers from one pick area. On-time shipping is orders shipped by the promised ship date divided by orders due to ship, with a target of 98%. Each late order means an expedite call from customer service and often premium freight.
2Current status
| Jan | 98.2% |
|---|---|
| Feb | 95.1% |
| Mar | 94.2% |
| Apr | 93.6% |
| May | 93.9% |
| Jun | 95.4% |
| Jul | 96.2% |
| Aug | 96.0% |
| Sep | 97.3% |
| Reason | Apr and May | Sep |
|---|---|---|
| Short at the pick face (stock in the system, not in the location) | Apr and May: 118 | Sep: 20 |
| Released too late to pick before the carrier pickup | Apr and May: 71 | Sep: 12 |
| Carrier missed the pickup | Apr and May: 31 | Sep: 14 |
| Pack or label error caught at the dock | Apr and May: 26 | Sep: 11 |
| Other | Apr and May: 16 | Sep: 4 |
| Total | Apr and May: 262 | Sep: 61 |
September: 2,250 orders, 61 late, 97.3% on time.
3Target
98% on-time shipping every month from December. At 2,250 orders a month that means 45 late orders or fewer (2,250 × 2% = 45). September had 61, so 16 more a month have to go.
4What we have done (PDCA)
Cycle 1: shortages (Apr to Aug)
- Plan
- Shortages were 45% of late orders (118 of 262). Cause: a location was refilled only after a picker found it empty.
- Do
- June: minimum levels for the 150 fastest-moving SKUs; the replenishment driver works from a minimum-level report every 2 hours.
- Check
- Shortage late orders fell from 59 a month (Apr and May) to 22 (Jun to Aug). On-time shipping rose to 95.4% to 96.2%, then went flat.
- Act
- Minimum levels made standard for those SKUs and reviewed monthly. The plateau sent us to the next cause.
Cycle 2: late release (Aug to Sep)
- Plan
- Next biggest cause. Orders went to the floor in one batch at 1 pm; the last ones could not be picked before the 3 pm carrier pickup.
- Do
- September: orders released in hourly waves from 8 am; the last wave at 1 pm is sized to finish by 2:30 pm.
- Check
- Late-release orders fell from 34 in August to 12 in September. On-time shipping 97.3%.
- Act
- Hourly waves kept; a wave board added at the dispatch desk.
What is left, what is next and what we need
5Remaining issues
- Carrier missed pickups: 14 in September. Cause not known yet; the logistics coordinator has a month of actual pickup times.
- Pack and label errors: 11 in September. Not yet analysed.
- Shortages still 20 a month. The 150 SKUs on minimum levels cover most picks; the next 100 by picks do not.
6Next steps (cycle 3)
| Action | Owner | Due |
|---|---|---|
| Review a month of pickup times with the carrier; agree a fixed pickup window | Owner: Logistics coordinator | Due: 31 Oct |
| Pareto of pack and label errors by type; 5 Whys on the top one | Owner: Shipping lead | Due: 31 Oct |
| Minimum levels for the next 100 SKUs by picks | Owner: Inventory analyst | Due: 15 Nov |
| Report at the November operations review | Owner: Warehouse manager | Due: Nov review |
7Forecast
If carrier misses halve (14 to 7), pack and label errors fall from 11 to 6 and shortages fall by 5, September's 61 late orders would have been 44 (61 − 7 − 5 − 5). That is 98.04% on time: at the target, with no margin. All three actions have to work.
8Help needed
Support from the operations director in the carrier review: the pickup window is in the carrier contract, which the warehouse does not own.
Coach's notes: why each box works
Why it works: The measure is defined in one sentence, with the target, so the trend that follows cannot be argued about.
The weak version: A status A3 that assumes the audience remembers the definition from April. Readers change, and so do definitions when nobody writes them down.
Why it works: The trend shows nine months, the target and when each change went in, so a reader sees cause and effect in one look. The reasons table compares the same categories before and now.
The weak version: Only this month's number, or a chart that starts in June to hide the drop. A status A3 is trusted when it shows the bad months too.
Why it works: Turning a percentage into a count of late orders makes the gap concrete: 16 orders a month is something a team can plan against.
The weak version: Restating the target without the gap. "Target 98%, actual 97.3%" sounds close; "16 more late orders a month to remove" tells you how much work is left.
Why it works: Each cycle reads as plan, do, check, act, and every check is a number against the cycle's own target cause. The plateau after cycle 1 is reported and used: it is why cycle 2 went after a different cause.
The weak version: A list of activities ("held kaizen, updated SOP, trained pickers") with no check. That is a diary, not a status; the reader cannot tell what worked.
Why it works: It says plainly what is not understood yet (carrier misses), instead of guessing a cause to make the status look further along.
The weak version: No remaining issues box, so the review assumes the problem is solved because the number went up.
Why it works: Each remaining issue has an action, one owner and a date, and two of the three actions are analysis first, not a fix picked in advance.
The weak version: "Continue to monitor." A status A3 with no next step tells the reader the owner has run out of ideas.
Why it works: The forecast is arithmetic anyone can redo, and it admits the plan only just reaches the target. That is the honest basis for the request that follows.
The weak version: "We expect to reach target by year end" with no numbers. The reader cannot tell whether the plan is enough.
Why it works: One specific ask, to a named role, with the reason. This is what a status review is for.
The weak version: No ask, or a vague one ("more resources"). If the owner needs nothing, the meeting could have been an email.
Common weak spots, and how to fix them
What reviewers find most often on first A3s. Use the list to check your own draft before the review.
Solution jumping
- Looks like:
- Theme: "Install a new lubricant dispenser on line 3." The rest of the sheet is written backwards to justify it.
- Why it hurts:
- The analysis can only confirm the answer already chosen, so the real cause (here, a change process with no standard work step) is never found and the problem comes back on the next line.
- Fix:
- Name the problem in the theme and keep countermeasures off the left side. Sobek and Jimmerson's rule: the theme should not advocate a particular solution.
Root cause is a person
- Looks like:
- "Root cause: operator did not apply enough lubricant. Countermeasure: retrain operators."
- Why it hurts:
- If every assembler on both shifts shows the same pattern, the cause is the method. Retraining people to work around a bad method lasts until the next new starter.
- Fix:
- Split the data by person and shift before naming a cause, and keep asking why until you reach something in the system: a standard, a tool, a check, a change process.
Goal not measurable
- Looks like:
- "Goal: improve changeover performance on line 2."
- Why it hurts:
- Nobody can say when the A3 is finished or whether it worked, so it never closes and nothing is learned from it.
- Fix:
- One measure, a number, a date and the data source. Add a second measure if the first does not by itself produce the benefit (changeover minutes and overtime hours).
No check of results
- Looks like:
- Follow-up box: "All actions complete." Ticks in the status column.
- Why it hurts:
- Finishing actions is not solving the problem. Without a measured result against the goal, a countermeasure that did nothing looks the same as one that worked.
- Fix:
- Write a prediction before the results come in, measure for long enough to see the pattern, and put the actual next to the prediction. A missed goal stays on the sheet with a next step.
Current condition with no numbers
- Looks like:
- "Leak failures are high and have got worse since the spring."
- Why it hurts:
- Without a baseline there is no goal to set and nothing to compare the result with. Opinions about the problem replace facts about it.
- Fix:
- Count it, over a stated period, and split it at least one way (by type, by time, by place). Then go and watch the work where the problem happens.
Savings that are never banked
- Looks like:
- "45 minutes saved × 6 a week × 48 weeks = 216 hours a year × full labour rate."
- Why it hurts:
- Freed hours are only worth money if a cost actually stops (overtime, a shift, a temporary worker) or the extra capacity is sold. Finance will see the gap.
- Fix:
- Value saved time at what really changes, cap it at what is spent today, and put the step that banks the saving (cutting Saturday hours) in the plan.
A status A3 that is a diary
- Looks like:
- A list of meetings held and actions taken, with this month's number and no trend.
- Why it hurts:
- The reader cannot tell what worked, what did not and whether the plan is enough to reach the target.
- Fix:
- Show the trend with the target and the dates of each change, one PDCA row per cycle with a measured check, and a forecast of what the next cycle has to remove.
Where the method comes from
The boxes and their order follow these sources. The examples, numbers and notes are ours.
- John Shook, Managing to Learn (Lean Enterprise Institute, 2008), chapter 1, pages 7 to 11
The A3 sections and the questions behind them (the Shook and Verble template), and the point that there is no single correct format, only the PDCA thinking behind it.
- Durward K. Sobek II and Art Smalley, Understanding A3 Thinking (Productivity Press, 2008), chapters 4 to 6
The three kinds of A3 (problem solving, proposal and status), each with its own chapter, and the A3 as part of Toyota's PDCA management system.
- Durward K. Sobek II and Cindy Jimmerson, A3 Reports: Tool for Process Improvement (IERC, 2004)
Theme names the problem, not a solution; current condition from direct observation; about five whys; countermeasures; a quantified prediction in the follow-up, compared with actual results.
Free templates and tools for your own A3
The blank A3, the root cause tools used in the examples, and calculators for the cost case.
- A3 problem solving templateThe standard seven-box A3 from background to follow-up, with a filled-in worked example. PDF and Excel.
- 5 Whys templateA problem statement, a why chain that can branch, evidence for every answer, root causes and countermeasures that are checked.
- Fishbone diagram template (6M)The same six-M fishbone on a landscape sheet for a session on paper, with a table for the causes you will check.
- Pareto chart templateA tally sheet, the Pareto by count and by cost with the cumulative line and the 80% cut, and a before and after comparison. Excel sorts and draws it.
- 4M change management sheet (change point)A shift log of every change to Man, Machine, Material, Method or Design, the extra checks after it and when normal checks resume, with a change point board.
- SMED changeover worksheetChangeover steps with times, internal or external, what can move, ideas and before/after totals.
- Cost-benefit analysis templateUp to three options side by side with costs, hard and soft benefits, payback, NPV and the benefit-cost ratio.
- PDCA templateOne improvement cycle on one page: plan, do, check against the goal, then standardize or adjust. PDF and Excel.
- Kaizen savings calculatorYearly savings, payback in months and first-year ROI of an improvement, from time saved, scrap avoided and its costs.
- SMED changeover calculatorTime saved by moving changeover steps from internal to external, capacity freed per week and the smaller batch it allows.
- First pass yield (FPY) calculatorFirst pass yield at each step and rolled throughput yield (RTY) across the whole process.
Go deeper
- A3 Problem Solving: A Practical Guide to Root Cause AnalysisWhat goes in each of the seven sections and how the A3 maps to PDCA.
- A3 Thinking: Beyond the TemplateWhy A3s written to fill the form fail, and how the coaching dialogue works.
- A3 Report as a Project Management Tool: From Problem to SolutionUsing the A3 to run and report a project, the status A3's home ground.
- Managing to Learn, John Shook (2008)How a mentor coaches an author through an A3, told as a story.
- Understanding A3 Thinking, Durward K. Sobek II and Art Smalley (2008)The three kinds of A3 and the seven elements of A3 thinking.
Doing this in LeanSuite
Continuous Improvement Software
LeanSuite's continuous improvement software runs every project the same way with built-in A3, DMAIC and PDCA templates, and tracks projected and actual savings for each one.
See Continuous Improvement SoftwareQuality Improvement
Log quality issues from the floor, trace each one to its root cause and follow the corrective action through to closure.
See Quality Improvement
FAQ
Questions about A3 examples
More examples
See all example galleries, including the DMAIC project examples, and the lean glossary.
Running A3s on paper and email?
See on a call how LeanSuite runs A3s from built-in templates and tracks projected and actual savings for each one.