An expense-approval tool for a 200-person scale-up, designed for managers who approve requests several times a day.
Approvals took 4 days on average, and 15% of requests got lost in Slack threads.
Fenwick grew from 60 to 200 people in 18 months, with teams in Stockholm, London and Berlin. Expense approvals still worked the way they did at 60 people: requests on Slack, shared spreadsheets, approvals by email.
A manager who sees "€212" on its own decides differently from one who sees "€212, and the team is already at 69% of its monthly budget". The request detail shows the team's remaining budget with a progress bar, plus the requester's history, so the decision happens in one place.
Why: it turns a mechanical approval into an informed decision, without opening another tool.
Approve, Request info and Reject are three fixed buttons, not a dropdown. "Request info" doesn't create a new screen: the request goes back into the same queue with a badge. Over-budget approvals don't close immediately; they move to a separate Finance queue for the final yes or no, and both the employee and the manager hear the outcome.
Why: for someone doing this five times a day, every hidden click adds up.
I ruled out two options: no indicator at all, and a full notification centre. The first risked missed updates for managers who open the tool rarely. The second would have turned Fenwick into a second Slack. A bell with a counter says "something changed" and nothing more.
Why: notifications stay where the team already talks; the action happens in the tool.
Three summary numbers at the top (waiting, total amount, over budget), then a table instead of cards, so amounts compare at a glance. "Urgent" and "Over budget" are always-visible filters, because they're what a manager checks first. The copy does its part too: an empty queue says "Queue clear", errors say exactly what went wrong, and dates are relative only within the last week.
Why: a power user needs speed on the hundredth visit, not guidance on the first.
A concept: targets, not measurements
The limit I'd fix first: the budget bar assumes the accounting system updates in real time. In a real company that takes weeks of work with engineering. Next time I'd ask the data team how realistic that number is before designing a whole screen around it.