The As-Is Situation
Our client, an international company of about 10,000 employees, has over 3,000 invoices to pay on a weekly basis to hundreds of providers.
Understanding the As-Is Process
The Accounts and Payments team is 10 persons, and they can do a quality and conformance check to about 20 to 25% of all payments. They give priority to amounts over 100K euros. As such every week for each country they will check if any issues:
- Check all invoices of 100K or above (these represent less than 5% of all payments to do)
- Check randomly about 20% of all invoices between 20K and 99K euros
- Ignore all checks of payments to do for invoices below 20K euros
So What’s the Main Problem?
The problem with such manual process is that almost 80% of the invoices are not checked before payment and these represent about 50% of total amount to pay – being several millions of euros weekly.
As-Is vs. To-Be Situation
Digitalys implements the To-Be Solution
We did an analysis of the current situation and created a to-be solution; in short, this is what we did:
- We defined business rules to use for all required checks
- We created a data process to give us each week the past 36 months of invoices and payments
- We automated all the business rules
- We elaborated an advanced algorithm that uses the historical data together with the business rules to find any potential issues in new payments for every week
- We developed an easy to use interface where Accounts and Payments team will only check those payments that have been determined by the algorithm as a potential issue
- We used machine learning to improve the algorithm taking into consideration the feedback when “payments with issues” were processed
- We reviewed the business process to fit in the new model.