A month-end close that took eleven days

The situation

Ridgeline is a regional bank with about 1,900 staff. Finance closed the books each month using a chain of extracts that had grown over fifteen years: eleven source queries, four spreadsheets, and two people who knew the order they had to run in.

The close took eleven working days. Worse, the figures did not match what the business units reported, so the first two days of every month went on arguing about which version was right.

Constraints

The bank is regulated, so every reported figure had to trace back to a source record, and no data could leave its own tenancy. Finance had no engineers of its own, so whatever we built had to be operable by analysts.

What we did

We modelled the general ledger into the warehouse first, then layered the business unit sources on top so both read from the same tables. Forty metrics that had existed in four versions were defined once in the semantic layer, each with a named owner.

Reconciliation moved from spreadsheets into tested transformations that run nightly. Where a figure disagrees with its source, the pipeline now raises it before anyone opens a report.

Results

Measure	Before	After
Working days to close	11	3
Manual hours per month	410	90
Metric definitions in use	4 versions	1
Time to answer an audit query	Days	Same day