Kadaster processes deeds - the legal documents behind every property transaction in the Netherlands. The team of clerks who handle this work each have daily targets. When the application slows them down, those targets are at risk.
And the application was slowing them down. Built top-down on the technical model - what the data structure could do - rather than bottom-up on how clerks actually move through a working day. So clerks built shadow systems: their own Excel sheets, notes on paper, screenshots in personal folders.
I joined as senior UX designer to redesign the core flow around the work the team was actually doing.
Three things kept coming up in interviews. Together they explained why the team's targets were under pressure - and why the official metrics never quite matched reality.
The app's screen flow followed the data model. The clerks' work followed a different sequence. Every act required navigating away from what they were doing to fetch context the system already had.
Clerks kept private spreadsheets to track which acts they'd touched, what they were waiting on, and what was flagged. The "official" tool was a dumb terminal for entering decisions made elsewhere.
Clerks were assessed on throughput. Some were under formal performance pressure. The tool that was supposed to enable them was actively making them slower.
Leadership wanted the team's role to move from data entry to quality control - but the tooling kept them doing entry. We had to make the controlling motion the easy one.
Sat with clerks while they worked. Recorded every detour: the Excel sheet they opened, the colleague they messaged, the note they scribbled.
Catalogued the shadow systems. Each one is a feature request the team had given up on getting from the official tool.
Redesigned the core screens around the clerk's actual sequence - not the data model's. Made the controlling motion the default action.
Low-fidelity prototypes back with the same clerks. Each round caught one or two new workarounds we'd missed.
Sanitised, but each one is a real thing a clerk said. They didn't need new features - they needed to stop fighting the ones they had.
"I keep my own Excel. It's the only way I know what I've finished today."
"My targets are public. I know I'm behind. I also know it isn't because I'm slow - the system is."
"To check one thing I have to leave the act, search a different module, come back, and hope I remember where I was."
"We're supposed to be the people who say yes or no. Right now we're typing."
"There's a shared sheet between us. It's not official. Without it I wouldn't know what my colleague is touching."
"I had a performance conversation about this. I was told to be faster. The tool was not part of the conversation."
A working set is in Figma. Public-safe screens will be added here as I get clearance.
All the context you needed two clicks away is now on one canvas. Decision panel on the right.
A clerk's day, their throughput, and what their colleagues are touching. Replaces both Excel sheets.
My contract at Kadaster was twelve months - at the end of it, I left to go freelance full-time. The new flow direction was set, tested in low-fi, and being picked up by the internal team for build-out.
What I'm proud of: how much of the team's quiet workaround economy we surfaced. Three shadow systems out of the dark in a few weeks of shadowing. That alone reframed the project for leadership.
What I'd do differently: spend more of the early weeks with line managers, not just clerks. The targets conversation was the load-bearing one - and I didn't get into it deep enough until late.