A professional-services group we worked with had reached the limit of an ageing on-premise ledger. The system still produced numbers, but it took a fortnight of manual work and a stack of spreadsheets to close each month, and nobody fully trusted the consolidated view. They wanted Oracle Fusion Cloud ERP, but the finance director was clear about the real constraint: the business could not afford a month where the books did not close.
That constraint shaped everything. A big-bang cutover would have put the entire close at risk on day one, with no clean way back. So we planned the move in stages, around the rhythm of the financial calendar rather than against it.
The test of a finance migration is not go-live day. It is the first close that has to land on a real deadline.
The challenges we had to solve
- A chart of accounts that had grown by accretion over a decade — duplicate cost centres, accounts that meant different things to different people, and no clean mapping to how the group actually reported.
- Opening balances and sub-ledger detail that had to tie back to the old system to the penny, or the new reports would never be believed.
- A close process that lived in the heads of two people and a set of spreadsheets, not in any documented procedure.
- A finance team that had been promised painless system projects before, and had reason to be sceptical.
How we approached it
We started with the chart of accounts, because everything downstream depends on it. Rather than lift the old structure into Fusion unchanged, we worked with the controllers to rationalise it — collapsing duplicates, agreeing what each segment was actually for, and mapping the old codes to the new ones so history could still be read. That mapping became the backbone of the data migration.
Then we ran a parallel close. For two cycles the team closed in both the old ledger and Fusion, and we reconciled the two line by line until the differences were explained or gone. It was deliberate, unglamorous work — but it is the only honest way to know the new system is right before you depend on it. By the time we cut over, the close had been documented properly for the first time, which mattered as much as the software.
Where it stands
The month-end now runs on one system the team trusts, in a fraction of the elapsed time it used to take, without the parallel spreadsheets. The first close on Fusion landed on its deadline — which, given how the project was framed, was the outcome that counted. The controllers can explain where every number comes from, and the close no longer depends on two people being at their desks.