Ellie

Context

Where it all started

This is what the team was working with every day: a colour-coded calendar, manually updated, shared across the dispatch team. It worked well enough when Thermondo was scheduling one visit per installation. Then the energy crisis hit, heat pump demand tripled, and suddenly every job needed three coordinated visits. The calendar couldn't hold it.

The Problem

How do you scale a Google Sheet?

The war in Ukraine changed everything overnight. Germany accelerated its shift away from gas, heat pump demand surged, and Thermondo's installation volume followed. What used to be a single site visit became three: a foundation visit, an electrical visit, and the heat pump installation itself.

The dispatch system was built for one. Nobody had planned for three.

The sheets were manually maintained, constantly out of sync, and completely unscalable. The backend platform existed but lacked the functionality the team actually needed, so the spreadsheet became the real system, and the platform became the backup. Tasks slipped through the cracks. Dispatchers worked blind. Field teams were sent out with the wrong skills, the wrong people, or incomplete information.

Research

You can't fix what you can't see

Dispatch sits at the centre of everything: the connective tissue between sales, technical surveys, field teams, and the customer. Before touching any interface, we needed to understand the full system: every handoff, every dependency, every place where something could quietly go wrong.

We ran four research streams in parallel:

Findings

Seen through the business lens

The business had a clear ambition: install more, get it right first time, and leave every customer happy. Simple enough on paper. But every one of those goals ran directly through dispatch, and dispatch was operating blind.

To scale installations, dispatchers needed to schedule faster and plan smarter. To improve first-time resolution, they needed to match the right technician to every job: the right skills, the right proximity, the right experience level. To protect CSAT, they needed to coordinate all three visits in a single conversation, not three separate ones weeks apart.

None of that was possible with seven Google Sheets and no shared visibility. The problem wasn't the calendar. It was that dispatchers had no system that let them do their job well.

Design for Impact

From calendar-first to project-first

The calendar was never the right mental model. It showed time slots, not jobs. It showed availability, not complexity. As long as dispatchers were thinking in calendar terms, they were going to keep making the same mistakes.

The shift we designed for was conceptual as much as it was functional: move from scheduling visits to managing installations. Every heat pump job became a single project object, containing all three visits, the assigned teams, the required tasks, the on-site dependencies, and the full status in one place. Dispatchers could finally see the whole job, not just the next appointment.

"As a dispatcher, I want to be able to plan all 3 visits at once so that I don't need to contact the customer 3 times."

The system introduced smart dispatching variables that had never existed before: proximity, skill set, and experience level. This gave dispatchers the information to make optimised decisions instead of educated guesses. The result was a direct hit on the two KPIs that mattered most:

Visuals

Ellie is Born

The platform launched with four core views, each designed to replace a specific failure point in the old spreadsheet-based process. Together they give every dispatcher, planner, and field team a single place to see, plan, and execute every installation from start to finish.

Challenges

Design didn't have a seat at the table, yet

We were the first product designers at Thermondo. Design wasn't in the vocabulary, and it certainly wasn't in the process. Decisions were made before we were in the room, and our job wasn't just to design Ellie. It was to prove that design was worth involving in the first place.

We did that by getting close to the people who mattered. Sitting with dispatchers. Joining operational stand-ups. Making our research visible so that everyone from QA to leadership could see the thinking, not just the output.

Alongside the product work, we ran a series of sessions to build design literacy across the business:

By the time Ellie launched, design had earned its place. Not because we asked for it. Because we showed what it could do.

Reflections

Learnings

Timing is everything: involve developers much earlier. Looking back, having an engineer involved early in the discovery phases would have saved us a lot of uncertainties and reduced the time spent refining tickets. Since then, we've made it a priority to include at least one front-end and one back-end developer in pre-refinement sessions, reviewing discovery findings, user interviews, and designs. This has made us more aligned as a team and significantly improved our delivery speed.

Less is more: MVPs reduce scope. Our ambition to solve as many user problems as possible led us to build a product that was too big for an agile team: technical debt, overloaded sprints with 100+ points, and growing frustration among developers. If I had the chance to design Ellie again, I would focus on a much smaller, more specific MVP that delivers value quickly while earning time and trust from the business and leadership.