Problem solving / RCA

A3 problem solving

A3 problem solving is licensed CC BY 4.0. Attribution: Katafacts (katafacts.com).

1 · What it is

What it is

An A3 is a single page that walks a problem from background through root cause to a tested countermeasure and its results — named after the paper size it was originally constrained to. The one-page limit is a discipline, not a formatting quirk: it forces you to separate what matters from what doesn't before anyone reads it. A properly built A3 isn't just prose — it carries a TAGS box (Target, Actual, Gap, Standard) tracking the metric that proves the problem is real and later proves it's fixed, and a Pareto breakdown that earns the right to chase one root cause instead of guessing at all of them.

2 · When to use it

When to use it — and when not to

Use it when

A specific, bounded problem with a measurable gap between current and target performance.
You can gather real data or direct observation about the current condition — not just an impression of it.
You want a record that survives past the meeting it was written in.

Not when

The "problem" is really a choice between options with no current-state gap to close — that calls for a decision tool, not an A3.
You don't have — and can't get — any real data or observation. Go and look first; don't fill the page in from memory.
The scope is bigger than one person or team can move in a matter of weeks — that's Hoshin-level work, not an A3.

3 · How to fill it in

How to fill it in

Background: State the problem in plain terms and say why it matters now — a metric that's off target, a complaint, a near-miss. One paragraph. If you can't say why this matters in one paragraph, you don't understand the problem well enough yet to write the rest of the page.
Problem statement: One or two sentences, separate from the background: the specific, bounded gap you're closing. Background is why it matters; the problem statement is exactly what's broken. If the two read identically, the problem statement is still too vague.
Target · Actual · Gap · Standard: Give the Standard (the baseline or spec this is measured against), the Target (the end-state goal), and a dated Actual series showing the trend. Actual and Gap should never be typed in by hand as static numbers — Actual is just the latest point in the series, and Gap is Target minus that point, so they stay correct as the series updates.
Pareto — where to look: Break the problem into categories and rank them by count, cost, or impact — pick the axis deliberately, because counting frequency when cost or customer impact is what actually matters is the single most common way Pareto analysis misleads. Chase the category that dominates, not the one that's easiest to fix. If that category is still broad, break it down one level further (L2) rather than moving on to the second-place category.
5-Whys — root cause: Start from the Pareto's vital-few category, not the problem in general, and ask why, repeatedly, following the strongest evidence at each step rather than the most convenient answer. Stop when you reach something you can actually act on — typically 3 to 5 whys deep. If the last why is "people weren't careful enough," you stopped one step too early.
Countermeasures: Each countermeasure must trace to the stated root cause, not just address the symptom. Name an owner and a due date for each — a countermeasure without an owner doesn't happen.
Effect confirmation: Decide before you act how and when you'll check the metric again — which cadence, over what period, against which of Target or Standard. This is where most A3s go soft: "monitor results" is not a confirmation plan.
Follow-up actions: Capture what has to happen to make the fix stick — standard work updates, training, wider rollout. This is what separates a one-time fix from a sustained improvement.

4 · What good looks like

What good looks like

The example below is a full worked A3: a TAGS box and trend chart showing a real, quantified gap; a Pareto breakdown — including a second-level stratification of the dominant category — that earns the root cause chain rather than assuming it; a causal chain that ends in something actionable; countermeasures that each name the cause they address; and a confirmation plan set before the fix went in.

Same example, as a downloadable xlsx workbook.

Download .xlsx

A3 — problem solving

Reducing late shipments from Packaging Line 2

Priya Nair, Line Supervisor · 2026-02-03

Team: Marcus Webb, Paint Booth Lead, Dana Ruiz, Scheduling · Sponsor: Oksana Petrova, Plant Manager

Background

Line 2 ships to three regional distribution centres on a fixed daily truck window. Two of those centres have raised late-shipment complaints in the last month, and the plan-of-the-day meeting has flagged Line 2 dispatch three times in the last two weeks. If the pattern continues, the next quarterly customer review will treat it as a service-level failure rather than a one-off.

Problem statement

Line 2's on-time shipment rate has fallen below its contractual 95% standard and is missing its 97% target.

Target · Actual · Gap · Standard

On-time shipment rate

Target

97%

Actual

81%

Gap

16%

Standard

95%

2025-12-082026-02-02

Pareto — where to look

Ranked by frequency

Waiting on parts from paint booth24 · cum. 57%
Truck arrival timing mismatch8 · cum. 76%
Packing station short-staffed6 · cum. 90%
Order paperwork incomplete3 · cum. 98%
Other / one-off1 · cum. 100%

Waiting on parts from paint booth — breakdown

Small-batch/custom color changeover17
Scheduled booth maintenance4
Upstream line stoppage3

5-Whys — root cause

Why 1The packing station is regularly waiting on parts from the paint booth — the largest single cause bucket, and mostly (17 of 24 waits) tied to small-batch/custom-color changeovers rather than maintenance or upstream stoppages.
Why 2Small-batch and custom-color orders are being inserted into the paint schedule as they arrive, each triggering its own changeover, rather than being batched into set changeover slots.
Why 3There is no agreed rule for how small-batch orders get sequenced into the paint schedule, so the scheduler decides case by case — and the case-by-case call is almost always to protect that order's promised date at the cost of an extra changeover.

Root cause: No standard rule exists for sequencing small-batch orders into the paint schedule, so each one causes an unplanned changeover that starves downstream packing of parts.

Countermeasures

Set a fixed small-batch sequencing rule: the paint booth runs standard colors continuously through the shift and holds all small-batch/custom orders for two dedicated changeover slots per shift.
Linked cause: No standard rule exists for sequencing small-batch orders into the paint schedule.
Owner: Priya Nair · Due: 2026-02-17 · Expected effect: Cuts unplanned paint-booth changeovers from ~3/shift to 2/shift, recovering roughly 45 minutes of packing input time per shift.
Add a visual queue board at the paint booth showing the two changeover slots and what's queued for each, so the scheduler has a rule to point to instead of making a judgement call under pressure.
Linked cause: No standard rule exists for sequencing small-batch orders into the paint schedule.
Owner: Marcus Webb · Due: 2026-02-17 · Expected effect: Makes the new rule visible and self-enforcing at the point of use, rather than dependent on the scheduler remembering it under pressure.

Effect confirmation

Track the on-time shipment rate weekly against the 97% target for 4 weeks after rollout; expect it back above the 95% standard by week 2 and at or above target by week 4. Track paint-booth changeovers per shift on the existing changeover log, watching specifically for small-batch changeovers settling at 2 dedicated slots per shift.

Follow-up actions

Update paint-booth standard work to document the two-slot sequencing rule.not-started
Train all paint-booth operators on the new sequencing rule and the queue board.not-started
Review results after 4 weeks; if the on-time rate holds, roll the same sequencing rule out to Line 1's paint booth.not-started

5 · Common mistakes

Common mistakes

Root cause is "not enough training" or "people weren't careful."
Almost always means the why-chain stopped one step early. Ask why the training or attention wasn't there in the first place — usually a system or standard-work gap, which is something you can actually fix.
The Pareto is ranked by frequency when cost or customer impact is what actually matters.
A high-frequency, low-cost category can dominate a frequency-ranked chart and pull attention away from the few events that are doing the real damage. Choose the axis on purpose, and say which one you chose.
Countermeasures aren't linked to a specific cause.
A grab-bag of good ideas with no causal link fixes the mood in the room, not the problem — and afterward you can't tell which one, if any, actually worked.
Current condition is written from memory instead of data or direct observation.
An A3 built on an impression inherits every bias in that impression. Go and see, or pull the data, before writing this section.
Target and Standard are treated as the same number.
Standard is the baseline or spec you're measured against; Target is the goal you're committing to. Collapsing them hides whether you're trying to get back to acceptable or push past it.

6 · What it connects to

What it connects to

upstream

Pareto chart
Feeds the A3's root cause section directly — it's what earns the right to chase one category instead of guessing. A standalone Pareto worksheet is useful when the category breakdown needs more room or more data than the A3 page allows.
5-Whys
The root cause analysis section runs a 5-Whys inline, starting from the Pareto's vital-few category. A standalone 5-Whys worksheet earns its keep when the causal chain needs more room than the A3 page allows.
Fishbone diagram
Use one first when there are several plausible cause categories to sort through before the Pareto data is clean enough to rank confidently.

downstream

Countermeasure matrix
When an A3 surfaces more candidate countermeasures than fit cleanly on one page, rank them by impact and effort there before committing.
Run / control chart
The TAGS trend chart is a lightweight run chart. Once a countermeasure is live, a full run or control chart tracks whether the new process is actually stable at the new level, not just briefly better.
PDCA record
An A3's effect-confirmation and follow-up sections are one turn of PDCA. Use a PDCA record to track further cycles if the first countermeasure doesn't fully close the gap.

7 · Where AI helps

Where AI helps

Judgement — stays yours

Deciding which root cause is real, not just plausible
Choosing the Pareto axis — frequency, cost, or impact — for this specific problem
Committing to a target condition and owning it
Judging whether an experiment's result actually closes the gap

Analysis — AI helps

Pressure-testing a why-chain for gaps or unjustified leaps
Flagging when a Pareto ranking looks misleading for the stated axis
Drafting a first-pass background and problem statement from raw notes or data
Checking that every countermeasure traces to a stated root cause

Drudgery — automated

Computing Gap from Target and the latest Actual, every time the series updates
Formatting the page and keeping it to one sheet
Rolling owner and due-date fields into a tracked action list
Exporting to xlsx (and pptx for report-outs) in the house format

9 · Rate this kata