Executive Summary
This process focuses on scraping invoice data, updating it, and uploading it to SAP. Our client provides the invoices to Jo in a shared folder. Jo reads each one depending on the title of the vendor in the invoice.
Client/Organisation Overview
This convenience retailer, founded in 2001, operates nearly 6,000 stores across nine countries and generates more than $24 billion in annual revenue. Every day, it receives a significant amount of invoices throughout its operations.
The Challenge
There was a significant backlog of invoices which needed to be cleared, and like many processes, automating with Jo means that the team are free to complete more challenging work, and Jo completes these tasks with speed and accuracy.
The Solution
Jo uses AI to scrape the invoice and then tries to map that to a mapping file. The mapping file contains additional details that Jo needs to input into SAP, i.e., certain numbers or ID codes. Jo reads the PDF, gets the data, and then logs into SAP and inputs the specific data that it needs to such as the total amount, invoice date, and tax codes. Jo then goes on to another page where the invoice is attached. These are all the steps for an invoice. Once complete, Jo will continue to do this for all the invoices. As a final check, once the invoice is scraped, Jo will add details to a memory file so when it processes each row, it will check the memory file to ensure that an invoice doesn’t already exist with those details. If Jo does find a duplicate, it will skip this file. Once the invoice is processed, it will be moved to a process folder to separate the processes that have been done and haven’t been done. Jo also attaches a confidence rating to show how confident it is that Jo has found a match in the mapping file. Anything below 85% will be skipped and flagged. Each process will attach an output log, providing a summary of what was successful, skipped, or failed. The output file will give more detail on why a row was skipped, i.e. skipped because the confidence rating was too low. There'll always be details as to why something hasn't been processed, and our client can find them and check it that way.
Implementation
Our client came to us with a rough idea of the process they wanted us to implement based on how they currently do things. Jo follows similar steps to how this was previously done manually. We mapped out the steps and got the details confirmed by our client. We tested the automation and showed our client the output from what Jo had scraped. Then our client checked SAP to make sure everything on the output log matched. This was verified once or twice and then our client confirmed they were happy to move to the live run. During implementation our client discovered that some invoices VAT was set to 7% and we were able to add an exception to the rule to accommodate this. Our approach is to be very flexible to best meet our clients needs.
Results
Jo successfully cleared a backlog of around 300 invoices and will continue working on invoice processing on an ad hoc basis.
Conclusion
Through the automation of invoices, we’ve been able to provide a flexible service to our client to assist them in easing the backlog of invoices. Jo completes these tasks with speed and accuracy and has liberated the team to work on more meaningful tasks.
