Expense claim reviews with 95% less manual processing
Employee reimbursements were prepared and checked in spreadsheets, by hand. A Python script and a dashboard replaced the manual steps and cut the processing effort by 95%.
From our own project and product work. Client names and identifying details are removed, and some figures are rounded.
The situation
A large building-materials maker reviewed employee reimbursements with manual data entry and separate spreadsheet analyses. Reports came late, and the reviewers spent most of their time preparing data, not reviewing it.
What we did
- Agree what matters. We analysed the claims data first in spreadsheets and KNIME, and agreed with the team which insights were worth running every time.
- Write it once. A Python script now does the whole preparation and analysis, from the raw claims to the exceptions.
- Show it. An interactive Power BI dashboard reads the script's output, so reviewers go straight to the claims that need a look.
What happened
| Measure | Result |
|---|---|
| Manual data processing | Down 95% |
| Review | From checking data to reviewing exceptions |
| Time to build | Three months |
What changed
- From preparing to reviewing. The hours go into the claims that matter.
- From spreadsheets to code. The same steps run the same way every cycle.
- From a monthly report to a dashboard. Anyone on the team can filter, drill down and follow up.
Products and services
The Audit Suite tests 100% of transactions, expense claims included. See audit effort and data and AI engineering.