Coventry Building Society

Coventry Building Society

Making User Personas Actionable Through AI

The Impact

Transformed unused persona data into an AI-powered conversational tool, enabling marketing and innovation teams to validate campaigns against user personas in real-time—reducing decision-making stress and improving campaign-audience fit.

The Challenge

Coventry Building Society had invested heavily in user research and persona development, but the data sat unused. Why? Staff needed to contact the data team for insights—a slow, cumbersome process that created a bottleneck.

The real problem wasn't lack of information; it was lack of accessible, actionable information at the point of decision-making.

Project scope: 1 week with internal PM and developer, plus 6 CBS team members across innovation, technology, marketing, and data science.

My Role

UX Designer, responsible for understanding workflows, facilitating discovery, and designing an intuitive interface for AI-powered persona interaction.

Key Design Decisions

1. Empathy Mapping to Find the "Moment of Need"

Ran an 8-person empathy mapping workshop to understand:

  • When staff needed persona insights

  • How they felt in those moments (rushed, uncertain, stressed)

  • Where information would be most valuable in their workflow

Impact: Identified that campaign validation was the critical moment—staff needed confidence their work would resonate with the target audience.

2. Gaming-Inspired Character Select Screen

Drew from the gaming background to create a familiar, approachable interface a character select screen where each persona was represented visually and could be "chosen" for conversation.

Impact: Reduced intimidation factor of AI technology; made interaction feel natural and engaging rather than technical.

3. Voice Note Interaction Pattern Over Text Chat

Leveraged learnings from previous conversational AI work (Portrait) and user testing insights: voice note style interactions clearly signaled when the AI's "turn" was complete, reducing confusion.

Technical challenge: The AI technology couldn't facilitate real-time conversation smoothly, creating awkward pauses and confusion.

Solution: Introduced status messages with human touch ("Thinking about your campaign...") to manage expectations and make delays feel intentional rather than broken.

Impact: Transformed a technical limitation into a more natural, less stressful experience.

The Process

Discovery Phase

  • 8-person empathy mapping workshop with CBS staff

  • Explored existing apps with similar conversational experiences

  • Mapped staff workflows to identify integration points

Design & Testing

  • Initial prototype with text-based chat interface

  • User testing with 3 CBS team members revealed technological limitations causing poor experience

  • Iteration #2: Added status messages and shifted to voice note pattern

  • Referenced learnings from other relevant projects on effective conversational AI patterns

Technical Constraints

  • AI couldn't support real-time conversational flow

  • Had to design around limitations while maintaining natural experience

Outcomes

Primary Goal: Reduce staff stress and uncertainty when creating campaigns

Secondary Goal: Improve campaign-audience resonance through persona-informed decision-making

Secondary Goal: Improve campaign-audience resonance through persona-informed decision-making

Key Learnings

Access Beats Perfection: The most comprehensive persona research is worthless if it's inaccessible when needed. Design for the moment of use.

Technology Limitations Require Creative UX: When tech can't deliver the ideal experience, thoughtful interaction design can bridge the gap. Status messages turned a bug into a feature.

Cross-Disciplinary Experience Compounds: Gaming background informed character select; Other project work informed conversational patterns; empathy mapping revealed the real problem. Every project can help build the next.

©2025 Jonny Gwillim. All rights reserved.