← All work

Case study

Turning dealer data and customer feedback into consistent coaching reports

An AI application that does the collation before a coaching visit, so coaches can spend their time on the coaching.

Dashboard of the AI coaching app listing a coach's dealer centres, with summary panels. Names and figures are anonymised.
The coach's dashboard. Client names, locations and all figures have been anonymised.

The challenge

Coaches for a global automotive manufacturer visit dealerships to help them improve performance. Before every visit, each coach spent about two hours collating dealer performance data and reading through thousands of verbatim customer survey comments. Every coach did this differently, so the quality and focus of coaching varied from one coach to the next.

My role

Project lead, named on the specification that goes to the client for sign-off. I designed the application with the client stakeholders who commissioned it. I was hired to produce e-learning content, and this project came about after I reshaped my role around generative AI; around 90% of my work is now AI-enabled job aids and performance support.

Constraints

  • Client confidentiality: no client data, individual names or system details can be shared.
  • Large volumes of unstructured data: thousands of free-text survey comments alongside performance figures.
  • The output had to be something coaches would trust and use, not just a summary.

Key decisions and why

  • Let the tool do the collation, not the coaching. The application produces a structured coaching report; the coach still brings the judgement and the conversation.
  • One structure for every report. A consistent format means every coach starts from the same picture, which tackles the variation between coaches directly.
  • Measure what the tool can actually change. I chose preparation time per visit and consistency between coaches as the measures, rather than dealer performance. That data comes from people I can reach, arrives in weeks rather than months, can be attributed to the tool, and leaves time to change course if it doesn't move.
  • Beta first. The application is in beta with the client's lead coaches before full release.

What we built

An application that takes in dealer performance data and customer survey comments and produces a structured coaching report for each visit.

A single dealer centre's page in the AI coaching app, with performance panels and themes drawn from customer comments. Names and figures are anonymised.
A dealer centre's page, the starting point for a coaching visit. Anonymised.

Results

  • Designed to replace about two hours of manual collation before every coaching visit.
  • Removes the variation in how individual coaches prepare.
  • Currently in beta with the client's lead coaches.

What I'd do differently

I designed with the client stakeholders who commissioned the tool. With more time, I would also have interviewed the coaches who use it every day, because they would have told me different things.

Let's talk

I'm open to new roles in AI adoption and enablement, AI-enabled learning and senior learning design, in London, the Cambridge area or remote.

The quickest way to reach me is a LinkedIn message or a connection request with a short note.

Connect on LinkedIn (opens in a new tab)