Process
Analysis & design The upstream work

Before the build

Every demo on this site was shaped before any authoring tool was opened. This page shows that upstream work: the analysis, the modality call, the SME questions, the evaluation plan, the storyboard, the journey map, and what the pilot changed. Each artifact is anchored to a project on this site you can go play.

Method · ADDIE, front-loaded
Lens · Human-centered design
Rule · Assessment before activities
ARTIFACT 01 — NEEDS ANALYSIS

The request was a video.

This is the condensed analysis brief behind The Return. The stakeholder asked for a video. The data asked for something else.

Play what it became →
ANALYSIS_BRIEF — RETURNS_TRAINING.DOC condensed
THE REQUEST
"Associates keep messing up returns. Can we get a training video before the holiday season?"
WHAT I ASKED FOR FIRST
The refund-error log, three shift observations, and ten minutes with two associates who "keep messing up."
WHAT THE DATA SAID
Errors clustered on no-receipt card purchases and condition codes — two decision points, not general ignorance. Associates could recite the policy; they froze applying it with a customer watching.
THE REFRAME
Not an information problem — a practice-under-pressure problem. A video adds information. Only rehearsal transfers to the register.
// DECISION: simulation with a coach that fades out. The second, unguided run is the assessment. Video request declined, with the error log as the reason — stakeholders accept "no" when it arrives with their own data.
ARTIFACT 02 — MODALITY CALL

Format follows constraint.

The decision matrix behind Cold Chain. The audience is deskless, mid-shift, and phone-first — three formats were on the table, and two of them lose to the constraints, not to taste.

Play what it became →
CONSTRAINT
45-MIN COURSE
WALL POSTER / JOB AID
SPACED MICRO ✓
No desk, no PC, 3-min breaks
Fails — needs a back office and a covered shift
Passes
Passes — phone, 3 min, twice a week
Numbers must survive 6 months
Fails — one exposure, then the forgetting curve
Partial — becomes wallpaper by week three
Passes — retrieval at widening intervals
Manager needs completion data
Passes
Fails — a poster reports nothing
Passes — per-item mastery, not seat time
// NOTE: the job aid still ships — as the in-app policy card, performance support at the moment of need. Modalities aren't rivals; they're layers. The matrix decides which one carries the learning.
ARTIFACT 03 — SME INTERVIEW GUIDE

The five questions I ask an SME first.

Five questions from the kickoff behind The Visitor, with why each one is asked. Experts narrate what people should know; these questions extract what people must do — and where it actually goes wrong.

Play what it became →
Q1
"Walk me through the last time this actually went wrong."
// Gets a story, not a policy recital. The Visitor's courier-with-full-hands is a real incident, lightly disguised.
Q2
"What does a smart, well-meaning person get wrong here?"
// Surfaces the plausible wrong answers. Every distractor in the scenario came from this question, which is why none of them feel dumb.
Q3
"If they remember only one thing at the door, what must it be?"
// Forces prioritization. The answer ("make verification the easiest path, not an accusation") became the course's spine.
Q4
"What do you wish people would stop doing, even though it feels helpful?"
// Finds the misconception worth un-teaching: confronting the intruder yourself. It became a quiz distractor with targeted feedback.
Q5
"How would you know, 90 days from now, that this training worked?"
// The evaluation plan starts in the kickoff, not after launch. The SME's answer — faster reporting, not fewer incidents — set the L4 metric.
ARTIFACT 04 — EVALUATION PLAN

Kirkpatrick, honestly practiced.

The four-level plan for Coach the Coach, including the level most plans quietly skip. Each level names its instrument, its timing, and its owner, because a metric nobody owns does not get collected.

Play what it became →
LEVEL
INSTRUMENT
WHAT IT PROVES
WHEN · WHO
L1 Reaction
Three questions max, asked at week 3 — not the end — so there's still time to fix the program.
Whether practice feels safe enough to be honest in. Nothing else.
Wk 3 · ID
L2 Learning
Scored role-play against the observable rubric, weeks 1 and 6. Same scenario difficulty, different scenario.
Skill gain a transcript can prove — not quiz recall about coaching.
Wk 1+6 · Facilitator
L3 Behavior
Two 10-minute floor observations per lead at 30 and 60 days, using a 5-item checklist their manager also holds.
Transfer. The level everyone cites and almost no one schedules — so it's scheduled before launch, in the managers' calendars.
Day 30+60 · Manager
L4 Results
Team retention and internal-promotion rate for coached teams vs. baseline, at 6 months.
Directional evidence only — and the plan says so out loud. Claiming causation from n=12 teams would be fiction.
Mo 6 · L&D lead
ARTIFACT 05 — STORYBOARD

Six frames before any tool opens.

The whiteboard storyboard behind The Visitor. Low fidelity on purpose — at this stage the SME argues with boxes and arrows, which is cheap. Nobody argues with a finished Storyline file, which is expensive.

Play what it became →
1. welcome — no objectives dump yet, just the hook
2. the 3 tactics as flip cards — learner reveals, not told
3. branch map — 3 doors, 7 endings, every path replayable
4. feedback = consequence, never "incorrect!" (see writing samples)
5. check: 2 Qs only — the scenario already assessed them
6. summary → "try another path" is the real CTA. replay = the lesson
ARTIFACT 06 — JOURNEY MAP

Ninety days, paced instead of poured.

A retail onboarding I mapped with backward design. New hires were overwhelmed for 30 days and undertrained for the next 60, so the sequence got three phases with manager touchpoints at every seam. The dotted line is the thing being designed: confidence.

WK 2 · "I'm drowning" — the dip the old plan ignored DAY 90 · runs a shift solo
DAYS 0–30 · SURVIVE
Only what today's shift needs.
Register basics, safety, who to ask. Everything else is deliberately withheld — the fire hose was the original problem.
DAYS 31–60 · GROW
Edge cases, with backup nearby.
Returns, escalations, the harder conversations — introduced once the basics are automatic and working memory is free for them.
DAYS 61–90 · OWN
Teach it back.
Shadowed by the next new hire. Explaining the job is the assessment, and it staffs the next cohort's support.
● MANAGER CHECKPOINTS: DAY 7 · 30 · 60 · 90 — 15 minutes, same 4 questions both sides know where they stand
ARTIFACT 07 — THE PILOT

What five learners changed.

Every module on this site gets tested with the people who'll use it before it ships. This is one change from the Cold Chain pilot, kept small on purpose — pilot findings usually are. The design isn't done when I like it; it's done when they stop tripping on it.

Play the shipped version →
BEFORE — DRAFT 2
"Cool food from 135°F to 70°F within 2 hours, then to 41°F within an additional 4 hours."
Accurate, and every pilot learner still answered the two-hour question wrong. They heard one six-hour window, not two clocks.
AFTER — SHIPPED
"Two stages. Two clocks. Two hours to seventy, four more to forty-one."
Same rule, restructured around the mental model that failed. The two-clock frame became the spine of the whole series, including the animation.
THE QUOTE THAT FORCED IT · PILOT SESSION 3 OF 5
"Wait — the four hours doesn't start until it hits seventy? Nobody's ever said that part out loud."
HOW THIS LOOKS ON A TEAM
Human-centered by default

Language tested with the people who'll use it, pilots before launch, and distractors drawn from real mistakes — empathize, prototype, test isn't a workshop poster to me; it's Artifacts 01–03.

Iterative, in the open

I work in drafts and reviews, not big reveals — storyboard, rough build, pilot, ship — with SMEs and stakeholders seeing work-in-progress at every step. Feedback early is cheap; feedback after launch is a redesign.

Independent, not siloed

Every artifact on this page is designed to be handed off — a matrix a Principal Designer can challenge, a rubric a facilitator can score with, a checklist a manager can hold. Work that only lives in my head doesn't count.