The Problem
The finance team of a UK professional services firm — four people supporting a business of around eighty — was running across four separate platforms. Invoicing happened in the accounting system. Expenses were captured by staff in a different mobile expense app. Reconciliation against the bank was done in the accounting system again, but with a separate plugin. And reporting to the leadership team lived in a recurring spreadsheet that the finance manager rebuilt each month by exporting from the other three systems.
Each platform did its job. The cost was in the gaps. An expense submitted through the mobile app had to be approved there, then re-keyed into the accounting system as a journal entry, then matched against a bank transaction during reconciliation. The same line item was touched four times across three systems. A single expense report could take twenty minutes of click-through across platforms to fully process.
The finance manager had quantified the drag and reported it bluntly to the leadership team: roughly one full day of every team member’s week was being spent on the boundary work between systems rather than on actual finance. The leadership team agreed something had to change but were nervous about ripping out platforms that individually worked well.
The Approach
We did not replace the four platforms. We built an integration and workflow layer that sits across them and removes the boundary work. Expenses approved in the mobile app push automatically into the accounting system with correct coding pre-applied based on a configurable rule set. Bank transactions land in the accounting system and are pre-matched against expected expenses and invoices, with the finance team’s role reduced to confirming the matches rather than making them. The monthly leadership report is generated automatically from the accounting system’s live data, with the rebuild step removed entirely.
The configurable rule set was the part the team owned. Expense coding had been a recurring source of small errors when done by hand, and codifying the rules — by claimant, by category, by amount threshold — let the finance manager encode their judgement once rather than reapply it every time. Exceptions still surface for human review; the rules only handle the cases that were always going to be coded the same way.
The System We Built
A workflow layer connecting the firm’s mobile expense app, accounting system, bank feed, and reporting layer via API. Configurable expense coding rules editable by the finance team. Pre-matched bank reconciliation with confirm-or-correct review. Automated monthly reporting pack generated from live data, replacing the manual spreadsheet rebuild. An exceptions queue surfacing anything the automation could not handle confidently — these are the cases the team’s judgement is needed for, and the team’s time goes to those rather than to routine processing.
The Outcome
The finance manager’s “day a week per person” measurement came down substantially. The team estimates that roughly half of the boundary work has gone and most of the rest has been compressed into the exceptions queue. They have used the freed time partly to clear a backlog of analysis the leadership team had wanted for two years and partly to take on the financial reporting for a recently acquired subsidiary without adding headcount.
The monthly report now lands in the leadership team’s inbox automatically on the first working day of the month, replacing a rebuild process that had been variably timed and occasionally late.
What We Learned
The “rip out and replace” instinct on a project like this is wrong. Each of the four platforms had been chosen for good reasons, the team knew them, and the cost of moving was not justified for any one of them in isolation. The work that was wasteful was the work between them, and that is the work software can absorb. Building the layer rather than replacing the platforms kept the team’s existing competence intact and got the gains faster.
Same Boundary-Work Problem in Your Finance Team?
If your finance team is spending more time moving data between platforms than analysing it, get in touch to talk through what a workflow layer could look like across your existing stack.