The Problem

When users signed up for the AI therapy program, they didn’t know what to expect.

They weren’t sure how to interact with the AI therapist, felt emotionally unsupported, and got frustrated quickly. Many of them dropped off right after their first interaction.

At the same time, internally, things weren’t easy either:

• The company was shifting from a waterfall culture to agile

• Product, design, and clinical science teams weren’t aligned

• There was confusion, tension, and different ways of working

  My Approach

  Having led through similar transitions before, I knew the problem wasn’t just about UX—it was about trust, understanding, and communication across the teams.

  1. Start with Listening

  I booked friendly one-on-one chats with key stakeholders.
  I wanted to know:

  • What are their goals?

  • What do they worry about?

  • How do they see UX and product supporting them?

  2. Build Shared Understanding

  I used visual storytelling and examples from my previous work to introduce:

  • Agile thinking

  • UX research workflows

  • Concepts like the Double Diamond, design sprints, and collaborative workshops

  This wasn't a formal presentation—it was a conversation.

  3. Run an Alignment Workshop

  Once we had shared language, I facilitated a kick-off workshop:

  • Used an Agile Research Canvas to gather evidence, hypotheses, goals, and questions

  • Co-created success measures and timelines

  • Introduced Assumption Smash and an Impact/Effort Matrix to align on the most urgent problems

  Problem Statement (Co-Created)

Users are dropping off early in the AI therapy program because their expectations of how it works are misaligned with reality.
During onboarding, they don’t get enough context. As a result, they feel confused, unsupported, and disengage quickly.

Solution: A Clearer, Kinder Onboarding Journey

Instead of trying to “fix everything,” we focused on expectation setting:

• Clear, honest guidance

• Reassuring tone of voice

• Light touch interactions that help users understand how to talk to the AI

I ran 3 rounds of usability testing and after each, I:

• Shared insights visually and clearly

• Referred back to our original workshop canvas to show how each insight tied back to our goals and KPIs

Business Objectives

1. Increase user retention during week 1

2. Reduce support tickets related to confusion

3. Define success for onboarding and improve completion rates

KPIs

• % of users who reported “not understanding” the AI

• Drop-off rate after the first interaction

• % of users who completed onboarding

• Time spent on onboarding (without early exits)

Results

• NPS increased

• Onboarding completion time was 3× faster

• fewer support tickets related to confusion