The interactive worksheet is loading. If this message stays visible in the Divi editor, open the page outside the Visual Builder to test it.

VISIBLE OR INVISIBLE

CHAPTER 15 · MAP BEFORE YOU AUTOMATE

A3 Workflow Template

A companion tool from Visible or Invisible by Sannah Vinding

Make one workflow visible. Agree on the problem, the evidence, and a small first test before adding AI support.

Where would you like to start?

Start a new workflow or return to one you saved. You can stop and save at any stage.

An A3 is a shared view of a problem and a proposed change. This simplified six-part worksheet brings the team’s thinking, ownership, and learning together.

When to use it

When a workflow has friction, unclear handoffs, or an opportunity for improvement.

What to bring

The people who do the work, a real example, and any baseline or source evidence.

What you’ll leave with

A shared picture, named owners, and a small test with an agreed review date.

Original web adaptation of the six-section structure in Chapter 15, Figure 19. The blue section is your current location, not a completion marker.

Your working copy

Build this A3 over several conversations. Complete what you know, note what is unknown, and save a working copy. You do not need to finish every section today. You can stop after Problem, Current State, or any section and return later.

Choose a size here, then match that paper size and Landscape in your browser’s print dialog. Use 100% scale, no margins, and no browser headers or footers. Black & white also converts images. These options apply only to printing; they do not change your working copy.

Save working copy keeps an editable JSON file, including images, for later. Print / save PDF makes a landscape overview with readable continuation pages when needed. Unfinished answers are welcome.

Session only · Your answers and images are not saved automatically.

How your working copy is kept

You can save an incomplete worksheet at any stage. This worksheet keeps answers in this open page by default. Closing or refreshing it can lose your answers and images. Images selected here stay in this page until you choose to save or download them; this worksheet does not upload them. This worksheet code has no analytics, AI service, or answer-submission endpoint. Other WordPress plugins or scripts on this page may have their own data practices; use fictional or non-confidential information while testing.

“Save in this browser” stores one copy only on this browser profile. Anyone with access to that profile may be able to open it. Clearing site data or using private browsing can remove it. It does not sync to another device.

Save working copy downloads an editable JSON file that you can reopen here later. Print / save PDF creates a snapshot for reading or sharing; it cannot be reopened as an editable worksheet. Use your own approved channel to share either file. Check that the file downloaded before closing. For team use, choose one facilitator to maintain the file; this is not live collaboration.

No saved browser copy found.

01

Problem

Describe the problem and its effect on the customer or team.

Describe one problem, its effect on the customer or team, and why it needs attention.

Name the workflow, product family, team, or customer situation you are working on now.

What are you deliberately setting aside for a later conversation or test?

02

Current State

Describe how the process works today, including where work gets delayed, repeated, or handed over.

List the steps, handoffs, delays, and workarounds.

What happens now? Include a measure, sample, or source. Mark estimates clearly.

Add a PNG or JPG from your device (up to 10 MB and 20 million pixels). It stays in this worksheet; no image is uploaded. The worksheet reduces the longest side to 1,600 pixels when needed and removes the original file metadata. Use a non-confidential image you have permission to share.

03

Root Cause

Look beyond the symptoms.

Write a hypothesis, not a conclusion you have not checked.

Note what you know, what is missing, and how you will check.

04

Future State

Describe how you want the process to work and what should improve.

Describe the change and where human judgment remains essential.

Use the same measure as your baseline where possible. Include quality as well as speed.

Add a PNG or JPG from your device (up to 10 MB and 20 million pixels). It stays in this worksheet; no image is uploaded. The worksheet reduces the longest side to 1,600 pixels when needed and removes the original file metadata. Use a non-confidential image you have permission to share.

05

Owners

Agree on the action and the person accountable. For Due date / checkpoint, use an exact date or a checkpoint such as “after source validation.” Include source and quality review ownership.

Ready to plan a test?

If not, save here and return when the team has enough clarity.

First Test

Plan the test

Keep the test within the scope agreed in Problem. Agree on evidence and boundaries before you begin, then come back to record what happened.

Name the scope, people or sample, timeframe, and review date.

Compare with the baseline and target. Set success criteria and a reason to stop.

What may AI prepare? Which approved sources may it use? What must stay out?

Who checks and approves? What needs escalation, and to whom?

When will you look at the evidence together? Leave blank until agreed.

After you’ve run the test

Come back here to record the actual outcome, learning, and your next decision. You can save before these answers are ready.

What happened? Compare the result with your baseline and target. Link or describe the evidence.

What changed your thinking? Note surprises, limitations, and open questions.

A person or the team makes this decision.

What will happen next? Who will do it, and when will you review again?

Make the work visible. Test the change. Keep people accountable.

The value is in the shared thinking. Review and revise this working copy with the people doing the work.

Fictional worked example

This made-up example shows one way to use the six sections. Names, dates, and results are fictional. It is read-only and does not change your worksheet. Close this example to return to your worksheet.

Workflow: Answer routine product specification questions
Worksheet owner: Maya, product operations
Date: October 5, 2026

01 / Problem

Sales representatives wait too long for routine product specification answers. Customers receive inconsistent response times, and the engineering team repeats the same searches.

02 / Current State

Current flow

A representative emails an engineer. The engineer finds a specification, checks the revision, and sends an answer. The answer stays in the email thread.

Baseline and evidence

A fictional review of 20 routine questions from the prior month found a median response time of two business days. Six requests needed a second search to confirm the current revision.

03 / Root Cause

Hypothesis

The team lacks one approved starting point for current specifications and a clear owner for checking sources.

Evidence and uncertainty

Five of the six repeat searches began with an outdated local file. We still need to check whether workload is also causing delays.

04 / Future State

Better flow

Route routine questions through a shared queue. Draft from a small, approved source set. An engineer checks every answer and source before a representative sends it.

Target

For the test sample, bring the median response time to one business day or less, with no incorrect specifications sent and a current source cited in every response.

05 / Owners

  • Identify and approve current specification documents. Who: Leo, engineering. When: October 7.
  • Set up the sample queue and record response times. Who: Maya, product operations. When: October 8.
  • Validate each answer before sending. Who: Leo, engineering. When: throughout the test.
  • Review evidence and make the next decision. Who: Maya and Leo. When: October 16.

06 / First Test

Small test and decision criteria

Use ten routine questions from one product family over October 8–15. Compare response time and source accuracy with the baseline. Stop immediately if an incorrect specification reaches a customer. Review together on October 16.

AI support and source boundary

AI may prepare a draft using only the approved public specification documents in the test source set. No customer identities, pricing, unpublished designs, or confidential files may be entered. The tool does not send a response or choose a source without review.

Human validation and escalation

Leo checks technical claims, revisions, and citations before a representative sends anything. Missing or conflicting source information goes to Leo; questions outside the product family leave the pilot queue for the normal process.

Actual outcome

The fictional test returned a median response time of 0.8 business days. All ten sent responses had a checked source and no incorrect specifications. Two drafts needed correction because similar product variants were confused; both were caught before sending.

Learning

The approved source set helped, but the product variant must be explicit before drafting. Ten questions are not enough to demonstrate reliable performance across the whole catalog.

Review decision: Adjust

Add a product-variant check and repeat the bounded test. Do not expand the scope yet.

Next step, owner, and date

Maya adds a required variant check to the team's process by October 19. Leo reviews the next ten questions on October 23.