Learning Experience Design · Scenario-based onboarding

The first fifteen minutes decide the whole shift.

Onboarding that trains last-mile delivery associates to decide under pressure — not to recite a checklist. Designed, art-directed and built end to end.

Play the scenario →Four decisions, two minutes, right here on this page.Read the design document ↓23-page PDF: brief, objectives, architecture, measurement.Open the storyboard →Ten annotated screens, developer-handoff standard.

Role: Sole designer and developer — brief, objectives, branching architecture, storyboard, art direction, Rise and Storyline build, accessibility, measurement plan
Client: MileOne Logistics — a realistic fictional carrier
Audience: New delivery associates and newly promoted dispatch coordinators
Duration: 37–48 min · Rise 19–24 · Storyline 8–12 · field guide 10–12
Tools: Storyline 360 · Rise 360 · Figma · Illustrator · Photoshop · Audition · NVDA and VoiceOver

MileOne Logistics is fictional, created for this project. All scenarios and data are illustrative. No metrics here are reported outcomes.

01The Problem

The gap isn't knowledge. It's deciding.

A last-mile operation loses more ground in the first fifteen minutes of a shift than in the eight hours that follow. Most new associates can recite the pre-shift checklist. What they can't do yet is decide under competing pressure — and that is the one thing a checklist cannot teach and a knowledge check cannot measure.

So the design problem wasn't "build a pre-shift course." It was: how do you let someone practise the first fifteen minutes of a bad morning forty times before it counts?

Fix

I can resolve this myself, right now, safely, within policy.

Flag

Someone else has the authority or the information. Tell them now, completely.

Note

This doesn't change today's plan, but it goes on the record.

Three dispositions, not two. A two-way model makes every right answer "escalate," which is useless as training and untrue to the job. The third creates the middle path where most real pre-shift discoveries actually live.

02The Scenario

Four decisions. Fifteen minutes. Try it.

A working miniature of the Storyline build — the same four decision points, the same three tracked dimensions, the same categorical-safety interrupt. Scoring and feedback are taken from the design document.

Safety70

Service70

Time left15 min

Decision 01 of 04 · 7:11 a.m. · Van 4412

Rear right is at 61. Spec is 80. Yesterday, same tire, you wrote down 78.

The station compressor is working and there's nobody waiting for it.

What do you do?

03How It's Built

A state machine wearing a story.

Branching: 38 slides across 7 scenes · 4 scored decision points · 12 tracked variables
Scoring: Points assigned per choice, so a native results slide reports to the LMS while custom Safety, Service and Time variables drive the coaching
Safety gate: Below 55, a Critical Risk layer interrupts, a two-screen micro-lesson runs, and the learner re-decides with a remediation flag set — so the LMS can tell a first-pass result from a second one
Consequence: Delayed on purpose. Top up the tire and go, and the warning returns at stop nine, on a bridge approach — then the station lead talks to you
Debrief: Replays the learner's path and shows how each choice moved each dimension. Fast-and-exposed gets different coaching from careful-and-costly, because those two people need opposite advice
Rise 360: Six lessons before the scenario for the model; a field guide after, built to reopen on a phone in a van at 6:55 a.m.
Accessibility: No timers anywhere · nothing signalled by colour alone · explicit focus order on every slide · captions and transcript · reduced motion set before the story starts
Standards: WCAG 2.1 AA · SCORM 1.2 and 2004 4th Edition · optional xAPI decision-path statements
Measurement: The scenario records a path string for every learner, so the plan surfaces which distractor a cohort picks most. That is a curriculum instrument, not a report

SB-06 · S3.4-DP1-CriticalRisk · one of ten storyboard screens

Layout: A LAYER, not a slide. Base dims to Ink at 85% over 300ms, no other motion. 2px Crimson #9E2B2Brule at y=180. "STOP." display serif 48px Paper at x=64 y=216.
Audio: One low sub-bass note, −26 LUFS, once, no tail. No voice — the absence of a narrator is more serious than a narrator.
Dev notes: Set three layer properties explicitly, because the defaults are wrong. On exit, apply the chosen option’s authored Safety cost — −20 here, −25 at DP4 — and clamp to 0–100. An earlier draft set varSafetyScore to a flat 60 instead, which meant a learner who arrived already below 60 gained points for committing the violation. A gate that can raise your score is not a gate. The re-trap that the flat value was there to prevent is handled by varRemediationFlag, which guards the floor trigger so it fires at most once per attempt.
A11y: The crimson rule is decorative; the word STOP. carries the meaning. Dismissible only via Continue; Escape must not close it.

04What I'd Do Next

Three honest next steps.

Test: Whether ninety-second triage survives real use. Five think-aloud sessions with associates inside their first three weeks, at a real station, before the shift.
Improve: Decision four is the weakest branch. Access notes vary too much by station for one scenario to carry — replace the single note with a pool of three, swapped by a random variable.
Scale: The dashboard, the three-dimension scoring, the remediation pattern and the debrief are a reusable shell. Swap the content layer and it's a warehouse shift start, a clinic opening, a field-service morning.

The instructional model was the hard part. The tooling wasn't. What I'd want a hiring manager to take from this is that I can be handed a messy operational problem, find the actual performance gap inside it, build a model simple enough to fit on a wallet card — and tell you plainly which parts still need testing.

Designing learning that holds up under real conditions.

Available for learning experience design and instructional design roles.

← All work