Ideally, every payment your business receives is neatly accompanied by the remittance advice which shares exactly which invoice is being paid and how much of it is being paid.
But according to the Federal Reserve Bank of MN and Remittance Coalition, about 88% of the remittance data sent by buyers is in a format requiring the seller to rekey data. That’s a lot of manual data entry as your AR team sorts through pdfs, word docs, excel spreadsheets, image scans, emails and even hand-written notes.
Does this sound more like your cash application process?
It’s frustrating, low-value but necessary work for your AR team, and it’s focused on shuffling papers and data, not growing your business.
But there’s still more layers of complexity to the cash application process:
- According to NACHA, 97% of businesses still pay at least some of their invoices with paper checks. This means there’s manual processing to receive, record, apply and recognize the revenue.
- Payments are coming from a variety of different sources in addition to checks, such as credit card payments, ACH or wire. And each of those channels has their own process for recording, applying and recognizing revenue.
- On the other side of these transactions is the buyer and their payment processes, which impact which credit card, card processors and fees might be applied, for example, and the time it takes to process the funds through those different channels.
This variability of payment channels and formats makes this part of the process one of the most time consuming and frustrating for AR team members, especially when we are talking about hundreds (or more) payments a day.
Why machine learning is “key”
Take “keying” or “data entry” out of the lingo of your cash application team, and they will thank you. Perhaps 100% elimination is not possible, but a significant reduction in low value labor and high intensity frustration is.
The two key steps that your team likely spends the most time on can be accelerated dramatically with a smart AR platform that leverages machine-learning technology.
- Remittance Advice Extraction: This is a machine learning-driven feature that scans inbound remittance advice emails and then pulls remittance information from those emails in a structured form. This often involves Object Character Recognition, which is extracting text data from images and PDF files. By intelligently identifying key words, account numbers and other data, it can then put that data into a structured format that is more easily accessible by your team - and is more readily available for the next feature.
- Remittance Advice Matching: Working with the structured data you just created from those messy email files in the previous step, this algorithm automatically matches the payments received to the correct open invoices on the account. This saves your team the time of digging through different remittance types and different funding sources to align the appropriate customer information.
The goal here is to process everything, but exceptions, which can then be processed by the members of the finance team. There will be exceptions, of course, but a good algorithm is looking to minimize those exceptions.
And because machine learning is an active technology, it’s ability to scan and process the data appropriately, in both steps, improves the more the algorithms are applied. So there will be fewer exceptions to manage over time, leaving your team more availability for high value, revenue generating activities. This also means you continue to increase in efficiency, even as your company - and payments - continue to grow: you scale without adding more headcount.
Smart AR platforms should be leveraging automation and machine learning across the credit-to-cash cycle in order to provide efficiency, cost-savings and a higher cash flow back to your business. If your current AR system isn’t doing that, it may be time to go shopping.
