11 to 2 days,
Rebuilding a lender's finance data platform and automating reconciliations
Data platform rebuild and automated reconciliations for a consumer lender.
The context
Month-end close took eleven working days. Finance pulled loan, payment and collections data from four systems into spreadsheets, then reconciled them against the general ledger by hand. Most of the time went on chasing breaks: payments that matched in one system and not another, often for reasons nobody had written down.
The data warehouse behind it was a decade old, loaded by overnight jobs that failed silently. Auditors had raised the manual reconciliation steps two years running. Finance wanted a faster close, but not at the cost of control: every adjustment still needed an owner and a reason.
What we did
- 1
Map the close
We spent the first three weeks with the finance team, mapping all 140 close tasks, their owners and dependencies. Reconciliations accounted for most of the elapsed time, so we started there rather than with reporting.
- 2
Rebuild the data platform
We replaced the overnight jobs with a Snowflake data platform built on dbt, with every model tested and documented. Loan, payment and ledger data now land daily with row count and freshness checks, and any failure alerts the data team.
- 3
Automate reconciliations
We encoded matching rules for each reconciliation, from bank statements to loan sub-ledgers. Matches clear automatically; breaks go to an exception queue with likely causes and the underlying records, assigned to a named owner in finance.
- 4
Reconcile every day
Because data and reconciliations now run daily, most breaks are found and fixed during the month. Month-end becomes a review of what remains, with a dashboard tracking every close task against its deadline.
The result
11 to 2 days.
Faster month-end close
Month-end close fell from 11 working days to 2, measured over the six closes after go-live against the six before. The finance team now signs off the books on the second working day, with far fewer late nights in the first week of each month.
Auditors closed their finding on manual reconciliations at the next year-end. Every adjustment carries an owner, a reason and a link to the source records, which made the audit itself faster for the finance team too.
- close tasks mapped and tracked
- 140
- of transactions matched automatically
- 92%
- consecutive closes signed off on day two
- 6
Have a number you need to move?
Tell us what it is. We will tell you what we would measure, and how long a discovery would take.