Mental Health Crisis Response Training
How I turned a failing product into Lumeto’s largest revenue generator in 6 months.
Company
Lumeto
Role
UX Designer
UX Researcher
Interaction Designer
Product Owner
Production Co-ordinator
Duration
6 months
Team
1 designer/PO (me)
3 engineers
2 technical artist
1 character artist
3 animators
Team
Clinical SMEs
Police training SMEs
Mocap, Voiceover, Production
The Problem
High Friction, Low Satisfaction
Behind Schedule
Reduction in Team Size
The Solution
Re-scoped and redesigned scenarios to significantly increase the number of available trainer actions.
Introduced key usability updates to the UI, resulting in a dramatic reduction in friction.
Identified and corrected UI elements creating unintended user behaviour.
Impact
⚠️
-70%
User Error
🏆
+200%
Scenario completion
😊
+3x
User satisfaction
🚀
Version 1.0
Successfully launched
Context
In early 2021, Lumeto partnered with the Ministry of the Solicitor General of Ontario to help reform police training, focusing on how officers respond to mental health crises. This involved translating six community-based de-escalation scenarios, previously performed with live actors, into immersive VR experiences.
Live-actor training, while effective, was costly and difficult to scale, making it impractical for widespread implementation. Each day of live-actor training could cost over $100k. VR offered a more scalable alternative.
Problem
Midway through development, three key team members left the project, causing major delays. The remaining team struggled to meet the project goals, and feedback from stakeholders revealed significant issues with usability, content accuracy, and overall satisfaction.
👥
50%
Team reduction
⏰
6 months
Behind schedule
😬
High
Levels of friction
👎
Low
User satisfaction
Discovery
User Research
To understand the issues firsthand, I organized user testing sessions with police officers at the Ontario Police College and in our studio. Users were asked to perform a series of unmoderated test scenarios, observed and recorded both in-and-out of VR. Follow-up surveys and interviews were conducted, where users could further contextualize their experience.
This revealed two major problems:
⚠️
The trainer’s UI was a significant source of friction.
💬
The content lacked depth and flexibility, leading to frequent scenario breakdowns.
Secondary Research
Collaborating with subject matter experts (SMEs) and academic partners, I reviewed training guides, curriculum materials, and live-actor training footage. This analysis confirmed that our content strategy needed a major overhaul to meet user needs.
Insights
Trainers often couldn’t find suitable responses to the learner’s actions, causing scenarios to stall. Additionally, the existing linear content model failed to provide the conversational depth needed to simulate real-life interactions.
Additionally, incremental development wasn’t viable due to limited resources and the inability to reshoot or record additional dialogue later in the process.
Exploration
User Journey
I identified a critical flaw in our initial approach: it was designed primarily from the perspective of the Person in Crisis. While consistent with the MHCRT curriculum facilitation guide, this perspective made it difficult for our team to accurately anticipate the needs of our users, the learners and trainers.
Redesign Strategy: I flipped the user journey to focus on the learner’s perspective and re-analyzed all previous research data through that lens. By mapping out the wide range of possible learner actions and aligning them with appropriate trainer responses, I created a more dynamic and flexible narrative framework.
Drag to explore – N+1 User Journey Map in FigJam
Expanded Narrative Framework
Rather than use a more typical linear branching structure for the user journey, I borrowed a spoke-and-hub structure more typically used in narrative design for the video game industry. This format better conceptualized the fluid and cyclical nature of the conversations we were looking to simulate, and ensured all possible actions from the learner had ample responses.
This approach increased the depth of interaction while minimizing additional production effort. Generic response options and reusable dialogue reduced the need for new content, ensuring no action led to a fail state.
UI Re-design
The original interface was adapted from a healthcare training design system, where precise monitoring of detailed tool use was crucial. However, in this context, the rapid pace of interactions required trainers to constantly shift focus between observing the learner and controlling the scenario.
The primary issue was tunnel vision—trainers were forced to divert most of their attention to a small portion of the screen displaying dialogue options. This led to a disjointed experience, making it difficult for trainers to maintain situational awareness.
Escalation
Additionally, the escalation feature—designed to reflect the learner's performance—caused unintended behavior. The feature allowed trainers to adjust a character's escalation level using a slider, with different levels unlocking new dialogue and actions. The intention was to encourage trainers to react dynamically by increasing or decreasing tension based on the learner’s actions.
However, due to limited response options, trainers began using the escalation slider as a workaround to access different responses. This led to constant toggling back and forth, which disrupted immersion and undermined the learning objectives.
Design Pillars
With our constraints and goals now clear, I worked with the product team to synthesize a set of four design pillars to guide our efforts. These pillars focused on reducing cognitive load, minimizing errors, and enhancing trainer efficiency while keeping development requirements manageable.
🤔
Reduce Cognitive Load
Ensure trainers could find and activate the correct response almost instantly to avoid breaking the learner’s immersion and maintain the flow of the scenario.
📉
Error Minimization
Design interactions to minimize the potential for errors, such as incorrect responses or delays, by providing clear visual and auditory feedback.
⚡️
Immediate Response Accessibility
Ensure trainers could find and activate the correct response almost instantly to avoid breaking the learner’s immersion and maintain the flow of the scenario.
🙌
Low Development Overhead
Create solutions that required minimal engineering effort, leveraging reusable assets and optimizing the existing system to deliver impactful results without extending timelines.
Final Design
Solution Elements
To address the UI friction, I implemented four key design changes:
💬
Contextual Grouping
⏰
Narrative Phases
🔁
Variable Actions
🎚️
Global Escalation
💬 Contextual Grouping
Grouped responses by conversation topics to streamline trainer navigation. For example, selecting "Drugs and Alcohol" would present relevant responses for that topic.
⏰ Narrative Phases
Segmented actions and dialogue into beginning, middle, and end phases to ensure appropriate options were available at each stage.
🔁 Variable Actions
Introduced multiple variations for common responses (e.g., "no," "stop") using pre-recorded alternative takes. This reduced repetition and enhanced immersion.
🎚️ Global Escalation
Kept all response options visible regardless of escalation level, with the slider controlling the intensity of the response. This improved discoverability and reduced trainer frustration.
Product Launch
We Shipped
The product was successfully deployed to all police services in Ontario. Within three months, over 1,000 officers were fully trained, and the ministry renewed its contract to expand the program.
Deployed to early adopters
We then proceeded to roll out our product to early adopter services, eventually deploying headsets to every police service in the Province, beginning with early adopters in and around the Greater Toronto Area. Processes I developed during user testing were used as a template to create in-person onboarding workshops where senior police trainers would participate in guided tutorials to familiarize themselves with the VR hardware, as well as our simulation platform. Upon implementation, user onboarding speed increased by X%.
Pilot study
We also worked with our academic partners and the ministry to conduct a pilot study into the efficacy of VR as a viable means of training for law enforcement officers, with their work appearing in the Journal of Police and Criminal Psychology, and Policing: A Journal of Policy and Practice.
Impact
And overall, with the changes in how we adapted the clinical content of the scenarios, and the design changes that were made to the trainer interface, we noticed 70% reduced error, 200% increased scenario completion, and 3x increased user satisfaction.
The product was successfully deployed to all police services in Ontario. Within three months, over 1,000 officers were fully trained, and the ministry renewed its contract to expand the program. Our VR training solution now serves over 18,000 officers across the province, offering a scalable, effective alternative to live-actor training.
⚠️
-70%
User Error
🏆
+200%
Scenario completion
😊
+3x
User satisfaction
🚀
Version 1.0
Successfully launched
💵
+$2M
Revenue
🏃♂️
+18k
Active users
👮♂️
1000+
Users fully trained
This project owes its success to the understanding and translation of it’s curriculum into a playable, intuitive, and accurate experience, along with a deep understanding of the needs of our users, and the ways they interacted with the product.
With so much work to get through in such a short time, and so few resources, there was little room for iteration once something made it into the code, so finding ways to increase confidence and reduce uncertainty in an extremely ambiguous problem space was crucial.
Testimonials
"Johnny’s can-do attitude and quick problem-solving make him indispensable. From untangling UX inconsistencies across platforms to handling engineering challenges with flexibility and composure, he consistently delivers polished, thoughtful solutions. Empathetic and collaborative, he fosters smooth decision-making and energizes those around him."
Belinda Darcy
Director of Design
"Johnny excels at transforming complex user problems into effective designs while proactively addressing challenges before they escalate. His curiosity drives impactful user discoveries, and his warmth energizes teams. A trusted leader and collaborator, Johnny consistently delivers meaningful contributions."
