Go Back

Mascot Design Case Studies

02
Prompt Engineering
Context Engineering
User Personas
Learning Experience
Compliance UX

Why Mascots? Because systems don't earn trust — characters do. Learning and compliance content at scale tends to feel cold, dense, and forgettable. A well-designed mascot turns abstract systems into a relationship — a guide the user recognizes, trusts, and actually remembers. This project produced two mascots for two distinct learning problems at Eli Lilly: DIGIT for data integrity, and Luma for the broader learning ecosystem. Both required tight prompt engineering to land the right visual, and tight context engineering to land the right voice across very different touchpoints.

2
Mascots Designed
50+
Character Variations
5
Learning Modalities
12
Weeks of Iteration

Two case studies in this set. One technical, one cultural. Both built character into systems that didn't have one.

CASE STUDY 01
DIGIT
Data Integrity & Governance Trainer
A friendly robot mascot designed to make ALCOA+ principles a daily practice, not a one-time module. Built for compliance teams, regulated-role onboarding, and SOP refreshers.
Compliance Robot mascot 30+ poses
CASE STUDY 02
LUMA
Learning Companion
The digital voice of learning at Lilly — a warm, approachable mascot who guides employees through training, onboarding, and continuous learning across global teams.
Cultural Human mascot 5 personality traits
Case Study 01

Meet Digit

Data Integrity & Governance Integration Trainer.

MEET THE MASCOT
Hi, I'm Digit.

I show up wherever ALCOA+ shows up. SOP refreshers, governance trainings, regulated-role onboarding, daily compliance nudges. I keep the principles short, the consequences clear, and the why never out of sight. I was iteratively prompted into being — fifty variations across pose, expression, accessory, and context — until trustworthy and approachable lived in the same character.

The Problem

Compliance training that's high-stakes and forgettable.

Data integrity training in pharma — particularly around ALCOA+ principles — is unforgiving. A single bad data entry can compromise patient safety, product quality, and regulatory standing. But the training itself? Dense, procedural, and easy to forget the moment the module closes.

STAKES
ALCOA+
Attributable · Legible · Contemporaneous · Original · Accurate · Complete · Consistent · Enduring · Available

Nine principles, one bad data entry, and the audit fails. Pharma compliance has no tolerance for forgetting.
MEMORY GAP
0%
of dense modules retained at week 3

Procedural slide decks score lowest on retention testing. Information without character doesn't stick.
OPPORTUNITY
5+
touchpoints where DIGIT can show up

Training modules, SOP refreshers, ALCOA+ nudges, governance reviews, regulated-role onboarding.
Approach

Four moves that built Digit.

From abstract compliance topic to companion who shows up across the workflow.

01
Anchor the purpose

DIGIT's job isn't to lecture — it's to make ALCOA+ feel like a daily practice, not a checkbox. Every interaction grounds back to: why does this matter to a patient downstream?

02
Engineer the visual

Iterative prompting to land trustworthy (lab coat), scientific (instruments visible), friendly (rounded silhouette), tech-forward (visor screen). Fifty rejections before five finals.

03
Write the voice

Direct, practical, lightly playful — but always credible. Always explains the why. Never punishes the mistake. "Let's catch this before audit" not "You forgot to log this."

04
Map the touchpoints

SOP refreshers, governance trainings, ALCOA+ nudges, regulated-role onboarding. The mascot becomes the mnemonic; the principles become the habit.

Variations

Thirty-plus poses, one character.

Each variation is the same DIGIT — different stance, expression, or context. Iterative prompting kept the silhouette consistent while flexing emotion and use case.

Voice

How Digit talks.

Direct. Practical. Lightly playful. Never punishes the mistake — always explains the why.

ON A MISSED LOG ENTRY
✓ DIGIT

"Let's catch this before audit. Add the timestamp and you're set."


✕ REJECTED

"You forgot to log this. Errors will be reported."

ON A SOP REFRESHER
✓ DIGIT

"Quick refresh on ALCOA+? It's been 90 days. Two minutes, then back to it."


✕ REJECTED

"Your SOP refresher is overdue. Please complete this module."

ON A VICTORY
✓ DIGIT

"Clean run. All nine principles hit. Audit-ready."


✕ REJECTED

"Module completed. Score: 100%."

Touchpoints

Where Digit shows up.

Five places across the learner's workflow. Same character, different moment.

01
SOP Refreshers

Quarterly nudges with a 2-minute refresh module

02
Governance Trainings

Lead instructor presence in interactive modules

03
ALCOA+ Daily Nudges

Single-tap reminders inside data-entry tools

04
Onboarding (Regulated Roles)

First face new compliance hires meet

05
Audit Readiness

Pre-audit walkthrough and confidence checks

From compliance topic to companionship.

DIGIT reframes a regulatory subject into an ongoing relationship. Learners aren't completing a module — they're meeting a guide who shows up across their workflow. The mascot becomes the mnemonic; the principles become the habit.

"

Every decision is only as good as the data behind it.

Case Study 02

Meet Luma

The digital voice of learning at Lilly.

Hi, I'm Luma. I'm the warm, approachable mascot for learning at Lilly. I show up wherever training, onboarding, and continuous learning happens — designed for inclusion, gender-neutral in tone, high-contrast for visibility, globally resonant. Five personality traits, deliberately balanced. The same Luma whether you're meeting me on day one of onboarding or in your tenth compliance refresher.

The Problem

Learning lived in silos. Each one had a different voice.

Learning at a global pharmaceutical company spans frontline manufacturing, lab science, corporate functions, and onboarding journeys across geographies. Without a unifying presence, every training feels like it lives in its own silo — different tones, different formats, no continuity. Employees needed a single human face of learning that felt supportive, not corporate.

MODALITIES
5+
training formats

Storyline, Rise, SimGate, video, microlearning — each previously had its own visual voice.
GEOGRAPHIES
30+
countries served

From frontline manufacturing in India to corporate functions in Europe and US.
AUDIENCES
4
learner segments

Manufacturing, lab science, corporate, onboarding — different needs, same Lilly.
CONTINUITY GAP
0
unifying presence

Pre-Luma, no consistent face anchored the learning ecosystem across teams.
Personality System

Four traits, deliberately balanced.

Designed to flex across modalities while staying recognizably her. Each trait checked against the others — no trait dominates, no trait disappears.

Supportive
Encourages without pressure
Uplifting
Small moments of encouragement
Relatable
Culturally neutral, warm tone
Human-like
Feels like a caring teammate
Where she lives

Wherever learning happens.

Designed for inclusion: gender-neutral in tone, high-contrast for accessibility, globally resonant.

01
Onboarding

Day-one greeting, navigation guide, role-specific paths.

02
Articulate Storyline

Interactive course narrator and progression coach.

03
Articulate Rise

Inline tooltip explainer and section-end recap voice.

04
SimGate

Lab-simulation companion for procedural micro-learning.

05
Video Modules

Animated presence in explainer and walkthrough videos.

06
Microlearning Nudges

Daily 30-second reminders embedded in workflows.

From icon to culture carrier.

Luma became a culture carrier, not just an icon. She represents Lilly's commitment to people-first learning, giving every training touchpoint a consistent face, voice, and emotional register. The result: learning that feels less like an obligation and more like a conversation.

Reflection

What I learned designing two characters.

01

Mascots are scaffolding for trust, not just personality.

Both DIGIT and Luma succeeded because their visual decisions traced back to a trust problem (compliance / cultural continuity), not a stylistic preference.

02

Prompt engineering ≠ prompt typing.

Fifty rejected variations to land five finals. The work wasn't in writing prompts — it was in defining the rejection criteria before generating anything.

03

Voice is harder than visual.

The mascot art was solved in 3 weeks. The voice took 6. Tone-of-voice testing across modalities is where the real product work lives.