7 Ways Machine Learning Enhances Purchase to Pay

By 2020, Gartner predicts that the use of artificial intelligence and machine learning will be pervasive.

In 1983, Sting famously released his song “Every Breath You Take” and sung “Every step you take, I'll be watching you.” And I’m not sure if the world has ever unanimously decided if we should be creeped out or feel safe when listening. In a lot of ways, I think that’s the association that a lot of people have with Machine Learning (ML) and Artificial Intelligence (AI), especially when it comes to using this technology in daily life.

But, there’s nothing to be afraid of. Better yet, Basware’s algorithm-based machine learning services are available to add value to your work at every point in your purchase-to-pay process by freeing up time, eliminating errors, and speeding things up.

Here are 7 real-world applications of ML and AI in purchase to pay:

1. Automatic Product Mapping for Fast Categorisation

It’s a beautiful thing when a buyer and supplier are on the same page. But, that doesn’t always happen. For instance, a product that fits under one category for the supplier, may be a totally different category for the buyer. If a supplier doesn’t know how to categorise their products so that it best translates for the buyer, machine learning can help fill in the gaps.
The Basware product mapping feature lets technology takes the wheel and organises products in the right category. Suppliers simply upload the catalog and products are then automatically matched to the buyer’s defined categories, driving consistency, making products easier to find in the catalog, and saving significant time.

2. Product Master Data Model for Simpler Catalogs

Some organisations have unique orders and need to make sure these products are available at a moment’s notice. This can put procurement professionals in tough situations as they are constantly having to negotiate different contracts with multiple suppliers for the same items to ensure availability of these products. Usually, this leaves the end user to compare multiple choices for the same product when all they care about is getting the product they need, and quickly.

Our e-procurement suite solves this dilemma. We use a flexible product master data model behind the scenes, so procurement departments can create a database of important products each with multiple suppliers, prices, units, etc., without subjecting the end user to all of those details. Then a highly intelligent algorithm automatically selects or suggests the best option for the user based on things like availability, location, and price of delivery. Machine learning like this promotes efficiency and makes ordering products simple.

3. Search Query Optimisation for an Improved User Experience

If you’ve ever searched for a specific product with no luck, you understand how frustrating this can be. Your search terms might make sense to you, but obviously not to the query. Basware understands the importance of search query optimisation to make the user ordering experience as fast and easy as possible.

We’re creating a link between queries that come up empty handed and the items that are then added to the shopping cart. In other words, if you search for “personal computer” and yield no results, but then later add a laptop to your shopping cart, our algorithm learns from these actions and gains a better understanding of what you were looking for initially. Pairing together the empty search results with the eventual shopping cart addition helps ensure there are fewer fruitless searches in the future.

4. Intelligent Order Aggregation for Increased Cost Savings

If you’ve ever shopped on Amazon, you know that at checkout you’re asked if you want your items shipped as they become available, or if you want to wait and ship your items in as few shipments as possible. If you’re eager to get the items quickly, you may choose to have them shipped as they become available, but if you want to save on shipping costs, you’ll opt to ship as much together as possible.

Within our e-procurement, we’ve recently debuted what we call “intelligent aggregation.” Our algorithm uses historical data based on past purchases to inform buyers that they may want to wait before issuing a purchase order (PO), as more items will likely be ordered from a particular supplier. Waiting to issue the PO saves you on shipping costs and may help you take advantage of volume discounts by ordering in bulk.

5. CloudScan for Efficiency Gains

For companies that prefer a DIY approach to invoice automation, Basware’s latest CloudScan features include the ability for customers to validate their scanned invoices by themselves or inject more automation into the process by having Basware do the validating. Once paper or PDF invoices are received, you can simply scan them using a compatible scanner. Then, you can choose if you want Basware to validate the information, if you want to use self-validation and do it yourself, or if you want to utilise a combination of these two options. Self-validation allows for the organisation to move invoices quickly into the workflow and through the payment process.

The intelligent part of this technology comes in the form of what CloudScan learns from your self-validation. Supervised machine learning algorithms absorb the data from user-made validation corrections and apply it to self-validations in the future. This expedites the process and helps eliminate potential human error. Learn how Basware customer, DHL, used Cloudscan among other Basware offerings to improve efficiency.

6. Smart Coding for Better Exception Handling

If there’s no purchase order (PO) created to match an invoice, an employee has to take the time and energy to manually code that invoice. But, with Smart Coding, our machine learning algorithm helps determine the code, freeing up that time for employees. The Smart Coding technology automatically searches and analyses your historical data and invoice coding templates to recommend the right general ledger (GL) coding of non-PO invoices. Using and learning from company financial data, the machine learning technology can continuously add data to its system, improving the accuracy of its recommendations.


7. Conversational Systems (Chatbots) to Speed-Up the P2P Process

Conversational systems, also known as Chatbots, are everywhere and their emergence the purchase-to-pay field was only a matter of time. Chatbots save time by allowing users to get a quick answer from their system instead of wasting time and playing the waiting game with back-and-forth correspondences.

A chatbot can answer many of the basic questions a user might have regarding the purchase-to-pay process. Users can use the chatbot to quickly find what they need, request new order requests from previous requests, and search the catalog to place an order quickly. Now the chatbot can also receive products into the system. And there’s plenty more. Chatbots help organisations increase efficiency and receive high levels of user adoption because of their easy interfaces and clear answers.

Ready to Learn More?

Embracing machine learning and artificial intelligence will add tremendous value to your organisation. Read our white paper 3 Disruptive Trends Shaping the Future of Finance and Procurement and, as always, contact us—we’re here to help!

Director of R&D