Ledger approvals
Cutting expense approval time in half for a spend management app used by 400 companies.
Role
Lead designer
Team
2 designers, 5 engineers, 1 PM
Timeline
7 months
Platform
Web and iOS

Context
Penwick helps finance teams at mid-size companies control what employees spend. Every expense over a set limit needs a manager's approval before it is paid back.
Note 1
I start every case with the money, because the people reading this care about whose money is waiting.
Problem
Approvals took 3.1 days on average. Managers approved in batches on Fridays, and employees waited a week to be paid back for work travel.
Note 2
Our first guess was that managers were lazy. Shadowing six of them showed they were afraid of approving something wrong.
Decision
We showed managers the three facts they checked by hand: the policy limit, the receipt match and the employee's past expenses. Clean expenses got a one-tap approve.
Note 3
The hard part was agreeing what 'clean' meant. Finance wrote the rule, I designed how it looks when the rule says no.
Outcome
Average approval time fell from 3.1 days to 1.4. Most clean expenses were approved on the day they were sent.
Note 4
The average looked good, so I went looking for the slowest expenses too. One had waited three weeks because a reminder never fired, and we fixed that as well.
What we cut
Bulk approve. Managers asked for it, and we built a prototype. In testing it brought back the Friday batches we were trying to end.
Note 5
Cutting a feature people asked for is the decision I'm proudest of here. I showed the test videos instead of arguing.
Before and after
Note 6
With the same screen size and the same data, the new queue puts the three facts managers checked right next to the amount.
Results
1.4 days
average approval time, down from 3.1
68%
of expenses approved the same day
22%
fewer expenses sent back to employees
Lessons
Managers approved faster once they trusted what the screen told them; removing taps helped less than we expected. Next time I would bring finance into the first round of research instead of the third.
Next project





