1 · Analyze — finding the question behind the request
Week 1 · inputs: quality records, near-miss reports, two dock observations · output: needs memo
Operations asked for "a short training about receiving chilled goods." The quality data said something sharper:
of 23 problem pallets last quarter, 43% were temperature abuse signed off on paperwork alone — competent people skipping
the probe step under time pressure. A content refresher would not touch that.
- The pivotal move: refusing to translate the request into topics. The real question became
"why do trained receivers skip verification when the dock clock is running — and what makes the right behavior the easy behavior?"
- Three failure patterns from records + observation: paperwork-only acceptance (43%), unnecessary defensive rejections (31%), dock-delay chain breaks (26%).
- Audience segmentation: new receivers (install the routine), experienced receivers (disrupt checklist blindness), shift leads (coach the borderline calls).
SME moment — the shift lead: the first SME conversation killed my draft objective. I had
written "receivers will understand cold-chain requirements." He said: "They understand it fine. They skip it anyway."
That one line moved the whole design from knowledge to decision practice — and produced the ≥60% abuse-discovery
reduction target, not a comprehension score.
2 · Design — action mapping before storyboarding
Week 1–2 · inputs: needs memo + SME triage · output: action map → storyboard → style guide
With the SME I ran a Cathy Moore action map: from the 90-day business goal down to three high-priority behaviors
and one supporting habit. Everything else — refrigeration theory, HACCP history, the full SOP library — was triaged out of
the module into job aids or the existing refresher.
- Triage test we used for each candidate behavior: "If a receiver skips this on a busy Friday dock, does a pallet go at risk?" Four behaviors passed. Fourteen content topics did not.
- Text-based storyboard with programming notes: 8 explainer scenes at second-level timing; module structure of four gates + three borderline incidents + certification bank.
- Mood board → style guide locked look before build: navy dock-world canvas, cool-blue interaction accent, green spent only on earned accept states, red reserved for real consequences.
Why scenario-based eLearning and not a classroom refresher: the failure is a decision under
pressure, not a knowledge gap. Learners need a place to make the wrong call and watch what it costs — a risk-free
dock where the worst outcome is a write-off that resets with a click.
3 · Develop — prototype first, then full build
Week 2–3 · inputs: storyboard + style guide · output: explainer + module + integration docs
The prototype covered Gate 2 and one borderline incident only — enough to test the interaction model
(choice → consequence screen → rule) with the SME before committing to the full build.
- Prototype feedback that changed the build: the SME flagged that my distractors were too obviously wrong.
Real docks aren't; the wrong choice is the reasonable-sounding one. I rebuilt every distractor from the failure
patterns in the quality records — "paperwork is complete, the driver is waiting" became the seductive wrong answer.
- Animated HTML explainer replaces video: 90 seconds, EN/NL subtitle toggle, WebAudio cues, no filming dependency — playable on the dock screen before shift.
- Full module: 21 decision points, every option wired to a consequence and a rule; 7-of-8 randomized certification bank at ≥85% pass gate (6/7 = 85.7%).
- Accessibility carried from day one: keyboard-complete, focus rings, reduced-motion support, no audio-dependent content, AAA/AA contrast per the style guide.
4 · Implement — deployment as designed behavior, not file handoff
Week 4 · inputs: build + pathway map · output: rollout package
Deployment goes through the moments where the behavior actually lives: team-lead toolbox talks before shift
(explainer), certification scheduled in week 2–3 of onboarding (module), and the decision-table job aid physically
at the goods-in desk — because a job aid beats recall at 6 a.m.
- Coach enablement: shift leads get facilitation notes for the borderline-call debriefs — the module certifies the decision; the lead calibrates the judgment on live loads.
- Pathway, not event: onboarding embedding, annual recertification, and escalation loop documented so the program survives its first staffing change.
5 · Evaluate — the plan was written before the module was
Day 0 → Day 90 · instrument: four-level measurement design
Kirkpatrick-style, pre-committed: L1 usefulness poll in-module (≥80% useful), L2 scenario certification (≥85% first
attempt) plus mentor sign-off for new hires, L3 observation cards on live borderline calls (≥90% full routine),
L4 the same numbers the action map promised — post-acceptance abuse discoveries ↓≥60%, unnecessary rejections <10%.
- Reporting rhythm: week-2 L1/L2 snapshot, day-45 interim trend readout, day-90 four-level report with a go/no-go on extending the format.
- Design-for-measurement: the module's item-level analytics report weak spots per gate — so iteration targets behaviors, not complaints.
Designed & built by I Wayan Usada