01 June 2020
Using data modeling to successfully extract, transform and analyze millions of data points to help detect fraud at a National Utility.
In this case study we will take a deep dive look at the successful completion of a fraud investigation at a large utility firm. The client runs its entire organisation on SAP and the key requirement for this project was a deep knowledge of SAP tables and business processes. As fraudulent activity was suspected among several senior executives, the investigation team had to operate discretely as part of the project remit.
Areas of focus included but was not limited to:
- Sales: Revenue assurance, customer master data validation, accounts
receivable process integrity checking
- Expenses: Procurement process validation, fraudulent transaction
reporting, vendor master data validation
- General data integrity checking and validation
- Client stock holding balances and analytical transaction processing
The client was in the business of holding inventories for many clients and there were concerns about the improper lending of inventories between
clients. To tackle this, the historic stock movement tables were recast to produce an hourly client balance for 6 years. The result was then scanned for unusual spikes and unexpected changes in stock balances (negative postings, movements with no purchase or sales document etc).
A once-off data extract from SAP and then modelled the various tables to perform several checks and answer questions. The scope of the project went back to the initial SAP implementation which resulted in more than 6 years of data and table sizes running into many millions of rows.