<< Back
Case Study on Using ML to Revolutionize Retail AP Operations

Case Study on Using ML to Revolutionize Retail AP Operations

August 25, 2023

Problem:

Market leader in retail and wholesale in North America, with a high volume Accounts Payable operation of >6M invoices a year of which 35% were non-EDI invoices totaling a volume of ~2.1M. The existing AP implementation had taken two years to get to a 49% straight though processing for non-EDI invoices, after starting at 4% straight through processing. Automation was therefore not yielding the required benefits, and there was a high manual intervention needed to resolve the 51% of unresolved invoices. Cost of manual processing was $4.5 per invoice against industry averages of $2.5 for Retail and Wholesale sector. Invoice cycle times and paid on time metrics were also compromised.

Solution:

Machine learning extraction and matching models were trained on 3 months of non-EDI data. The system was implemented in shadow mode so as to not disrupt the existing pipeline and to provide a comparative view of the efficiency of the heuristic system vs the machine learning system.

Outcomes: 

Against an existing straight through processing efficiency of 49%, Sapper APaaS was able to deliver straight through processing of 97%. Cost of manual processing was reduced from $4.5 to $2.5 due to automation of extraction and lesser exceptions. This savings of upwards of $6M dollars in manual processing costs, and vastly improved cycle times.