The challenge
Each day, one person downloaded payment files from several banks along with the open receivables report, then searched by hand for matches. The marked-up file went to a core team of 5 to 15 people, who looked for further matches. Fewer than 20% of payments were matched with confidence, and those were keyed into the ERP by hand.
Everything else went out by email. For every unmatched item, often hundreds a day, the team made a PDF of the details, attached them all to an email with the spreadsheet, and sent it to more than 175 people. Someone then watched the replies and copied each answer into a master spreadsheet so the core team could key it into the ERP.
The process used more than 60 hours of staff time every day, or over 15,600 hours a year. It also sent large amounts of payment data around by email. Slow cash application held up revenue recognition and incentive payouts to employees, and left reporting out of date.
What we built
Staff still download the bank files, but they now drop them in a shared location where an automation picks them up. The automation combines rules-based matching, machine learning, and AI to match 63% of payments, which then go to the core team for processing.
For unmatched items, the automation creates the PDFs the business relies on and stores them on an internal SharePoint site. It then sends the 175 recipients a single spreadsheet with the original payment details, suggested matches and confidence levels, and a link from each row to its PDF. One link lets each person claim their items.
The core team no longer sends emails, reads replies, or updates the master spreadsheet. They track every item from a single Power BI dashboard.
The results
The share of payments matched automatically rose from under 20% to 63%, and cash is now applied more than a day sooner. That speeds up incentive payouts to employees and gives the business more accurate reporting.
The core team’s daily workload dropped by 50 hours, about 13,000 hours a year of time returned to the business. Payment details now sit on an internal site with links to them, in place of hundreds of email attachments.