FINTECH, GLOBAL ENTERPRISE
Close at 1am Every Month. Until the Upload Loop Was Eliminated.
1am → daytime
month-end close transformation
10 iterations
upload loop eliminated
3 sets of books
unified in one validated model
The Problem
The company was a global FinTech running on a legacy Dynamics 2002 ERP with TM1 for consolidation and reporting. The month-end close routinely ran until 1am because the upload process between Dynamics and TM1 required up to ten manual iterations to balance.
Each iteration involved exporting trial balance data from Dynamics, formatting it for TM1, uploading, running the consolidation, checking for variances, and then going back to Dynamics to investigate and correct. The team was managing three sets of books across entities that had to reconcile before the consolidated numbers could be finalized.
The process was not broken in the sense that it always eventually worked. But it took an unsustainable amount of time and manual intervention, and the risk of error compounded with every iteration.
The Outcome
Novexa rebuilt the data flow between Dynamics and TM1 with a validation layer that caught discrepancies before the upload rather than after. The ten-iteration upload loop was eliminated. Data was validated at extraction, transformed with documented logic, and uploaded to TM1 in a single clean pass.
The month-end close moved from 1am to daytime hours. Three sets of books were unified in one validated model with documented intercompany elimination logic. The finance team stopped spending close night debugging upload errors and started closing with confidence.
The CFO described the change as moving from a recovery process to a confirmation process.
The Approach
- 1.
Mapped the end-to-end close process
Documented every step from Dynamics trial balance extraction through TM1 upload, consolidation, and variance checking. Identified the specific points where data broke and manual intervention was required. Quantified the average number of upload iterations per close.
- 2.
Built a pre-upload validation layer
Designed a Power BI validation model that sat between Dynamics and TM1. The model checked trial balance completeness, intercompany matching, and entity-level balances before any data touched TM1. Discrepancies were flagged and resolved before upload rather than discovered after consolidation.
- 3.
Unified three sets of books
Rebuilt the consolidation logic to handle all three entity books in a single documented model. Intercompany eliminations were automated with defined rules rather than manual journal entries. Currency translation was standardized.
- 4.
Redesigned the close workflow
Replaced the iterative upload-check-fix loop with a linear process: extract, validate, transform, upload, confirm. Each step had documented inputs and outputs. The close became a sequence of confirmations rather than a cycle of corrections.
A close process that ends at 1am is not a close process. It is a recovery process. When the validation happens before the upload, the close is a confirmation of work already done rather than a search for what went wrong.
Identifying details have been generalized to protect client privacy. Outcomes reflect actual engagement results.
Recognize this situation? Let's talk.
Book a Discovery Call