AI: Data cleansing & AI project
Transforming the finance function
SI
Services/ Goods
Background and objectives mission
A family-owned group was about to deploy a new P2P tool. The existing supplier repository, derived from several ERPs, had deteriorated (duplicates, data entry errors, incomplete data). The objective was to massively clean up this repository before the transition: guarantee the integrity of supplier liabilities, improve master data quality and prepare for electronic invoicing on a reliable basis.
approach adopted
We developed artificial intelligence scripts to analyze existing invoices and extract relevant supplier information. This approach automatically identified and corrected input errors, normalized data and removed duplicates. A de-duplication algorithm then merged similar records (followed by a final manual check to adjust ambiguous cases).
Benefits obtained
- The clean-up process has thoroughly cleaned up the repository: critical duplicates have been eliminated
- The new verified repository has enabled the P2P tool to be migrated without incident: master data error rate close to 0%.
- This preparatory cleaning has facilitated the preparation of electronic invoicing and offers total reliability of the supplier repository, which constitutes a "trusted repository", reducing accounting bottlenecks and improving overall team productivity.