“Just show me one project from start to finish.” It’s the request behind most questions about action mapping, and it’s a fair one. The model clicks faster when you watch it work than when you read about it.
So here’s one complete action mapping example: a fictional retailer, a measurable business goal, the analysis that followed, and the practice activities that came out of it, including the parts that deliberately became a job aid instead of a course. If you want the theory first, read the action mapping overview and come back. This page picks up where the diagram stops.
Meet Bluepine Outfitters
Bluepine Outfitters is invented. Any resemblance to a real retailer is coincidence, but the problem will be familiar to anyone in ecommerce. Bluepine sells outdoor gear online, and returns are eating the margin on footwear.
Maya, the operations director, sends L&D a familiar request:
“We need a course on the new boot fit guide and the returns policy for the support team. Can you have something ready by October?”
Buried in that request is a real business problem. Bluepine’s numbers (also invented, but plausible) say 28% of online footwear orders come back, most of them citing size or fit. Customers who chat with a rep before buying return far less often. There’s the opening a course factory would miss.
One thing before we start: action mapping isn’t a solo activity. Maya stays in the room, and so does Priya, a veteran support rep who acts as our subject matter expert. Everything below is decided with them, not handed to them.
1. Set a measurable business goal
The request mentions two documents and a deadline, but no goal. So the first conversation with Maya isn’t about content. It’s about what changes.
A weak goal would be:
“Teach the support team the new fit guide.”
Nobody can measure that, and nothing in it would move Bluepine’s return rate. Here’s the goal we write with Maya instead:
“Reduce the return-for-refund rate on online footwear orders from 28% to 20% by March 31, 2027, as support reps help customers choose the right size and fit before they buy.”
Every clause earns its place. There’s a measure (return rate on footwear), a baseline and a target (28% to 20%), a deadline, and the crucial “as” clause that ties the number to observable behavior in a specific job. If rep behavior doesn’t change, the number won’t move, and we’ll know.
Two things happen the moment this goal exists. Maya publicly owns the result with us, which makes the project accountable rather than decorative. And every later decision (what goes on the map, what becomes an activity, what becomes a job aid) has a test to pass: does it help move that number?
2. List what reps need to do, not what they need to know
Next question for Priya: what do reps need to do, on chats, to bring returns down? Not understand. Not know. Do. An hour with her and a pile of chat transcripts produces five actions:
- Ask what the customer plans to use the boots for before recommending a model
- Use each brand’s fit notes when advising a customer who’s between sizes
- Walk between-sizes customers through the printable measuring guide
- Offer an exchange instead of a refund when fit turns out to be the problem — every exchange is a refund that never happens
- Flag product pages where customer reviews contradict the stated sizing
Each action is observable. You could read a chat transcript and check whether it happened. The brainstorm also produced the usual impostors (“understand the fit guide,” “know the returns policy”), and we set those aside. Knowledge can support an action, but it isn’t one.
Just as important, every action links straight to the goal. An action map isn’t a mind map where anything vaguely related gets a branch. If a proposed action wouldn’t plausibly move the return rate, it doesn’t go on. “Learn the company history” never stood a chance.
The goal sits at the center of the map; the five actions branch off it, with sub-actions and notes hanging off the branches. It gets as complicated as the problem needs, and no more.
3. Find out why they aren’t doing it
Reps aren’t already doing these five things. Why not? Skipping that question is how you end up building a course for a problem a course can’t fix, so we run each action through the Will training help? flowchart. For Bluepine, that means interviews with six reps and a week of reading transcripts.
The environment was working against them
Two findings, neither of them a training problem. First, reps are scored on average handle time, and fit conversations take minutes. The scorecard was quietly punishing the exact behavior the goal needs. Maya agrees to add a quality measure alongside handle time. That’s a management decision, not a course.
Second, the brand fit notes live in a 40-page PDF that nobody can search mid-chat. The fix is to embed the notes in the chat console, two clicks away. Again: a tool change, not training.
Some knowledge belongs on a job aid
Priya points out that two fit rules answer most sizing questions: how each brand runs, and what to do when a customer is between sizes. Do reps need those rules memorized? No. They need them at their fingertips. So this part becomes a job aid, not a course: a one-page panel pinned inside the chat console.
The flowchart asks one more thing here: will people need practice using the job aid? For Bluepine, yes. Finding the right line mid-chat, under time pressure, is itself a skill, so the job aid shows up inside the practice activities below.
The rest was skill, plus a missing why
The between-sizes conversation is judgment, not recall. Reps get better at it by doing it, which means practice activities, not presentations. And one more finding: reps never see what happens after a fit recommendation goes wrong. The return arrives weeks later, in another system. A scenario can close that gap by showing consequences, and Maya can close it on day one by telling the team why the goal exists.
A note on starting points: new hires and five-year veterans don’t need identical practice. The activities below include easier and harder paths so experienced reps can skip ahead. One size never fit anyone, which is kind of the point.
4. Design practice activities, not a presentation
Here’s what the “training” actually looks like.
Activity 1: the chat lab. Reps work through simulated customer chats with the job aid open, exactly as they would live. One customer mentions wide feet and a backpacking trip; another loves a boot that runs narrow. Feedback shows the consequence of each choice, including the message that arrives “two weeks later” when the boots rub raw on the trail.
Activity 2: a branching scenario, “The Weekend Hiker.” The harder version of the conversation, where the customer is in a hurry and the popular model is out of stock. Choices branch, and the story plays out over weeks of compressed time, so reps finally see the return that a rushed recommendation causes.
Activity 3: spaced follow-ups. One short challenge a week for a month, built from real anonymized transcripts, so the skill doesn’t evaporate after launch week.
Notice what’s missing: slides, modules, a course wrapper. The only information in the whole design is the job aid and a few optional links inside the activities. People pull information at the moment they need it; nothing is pushed at them just in case.
“But where’s the course?”
There isn’t one, and Bluepine is better off for it. The activities live where the work lives: the chat lab runs in team meetings, the scenario lives in the console, the job aid is pinned next to the queue. A course is for when people need structured practice away from the workflow. That wasn’t this problem.
5. Decide what won’t become a course
The original request asked for a course on the fit guide and the returns policy. Here’s where each piece of that request actually landed:
- The fit rules became a job aid, tested with reps, pinned in the console, and used inside every activity.
- The returns policy became a searchable knowledge base article owned by operations. Policies change; a course would freeze today’s version and quietly rot. The article is open at the moment of need anyway.
- The handle-time scorecard became a management decision plus an announcement. Telling people about a new KPI takes a meeting, not a module.
- The product-page flags became a weekly report to the web team. Reps don’t need training to report a contradiction; they need somewhere to send it.
“But shouldn’t everyone at least read the policy?”
Reading isn’t doing, and nobody’s return rate moves because a rep once read a policy. The policy matters in the thirty seconds when a customer asks for a refund, and the knowledge base is right there. Exposure to information is how we got the problem in the first place.
What happens after the map
Maya’s dashboard now tracks the footwear return rate monthly, with a check-in planned for January. If the number hasn’t moved, we don’t blame the learners. We revisit the analysis. Maybe the environment shifted again, maybe we aimed at the wrong audience, maybe the goal needs adjusting. The map makes those conversations possible, because every piece of the design publicly answers to one number.
The activities iterate too. Chats that still go wrong become next month’s practice material.
Steal this approach
The request was a course. The course turned out to be the smallest part of the solution: a job aid, a couple of environmental fixes, and some focused practice did the heavy lifting. That’s not a failure of training design. That’s what training design is for.
For more worked material, the action mapping headquarters collects the overview, the flowchart, and the FAQs, and the book Map It walks through the full process with many more examples.
And if the branching scenario in step 4 is the part you want to build next, the scenario design toolkit lets you practice exactly that. You design scenarios for a fictional client of its own, with interactive worksheets for the tricky decisions.