Augmented Analytics in Accounts Receivable: Addressing Overdue Payments
When looking at how IT can empower an accounts receivable (AR) user or department to minimize overdue payments, the key metric for consideration is days sales outstanding (DSO). DSO measures the time it takes to collect payment after a sale is made. A low DSO (for example, 30 to 60 days) is good, indicating positive cash flow, whereas a high DSO (for example, 90 to 120 days) is not. It can indicate any number of issues, such as declining overall customer satisfaction or a need to extend credit in appropriate circumstances. Regardless, DSO is an important evaluation metric for shareholders and investors, and a high DSO can represent a red flag.
Traditionally, DSO is calculated on a monthly, quarterly, or annual basis. For the AR team to stay on top of cash flow, they need not only real-time visibility into overdue payments, but also predictive capabilities to forecast them – and, where possible, to take remedial actions before they become a bigger issue.
By supporting and providing augmented analytics, IT can provide collections managers with a view of all the critical measures they need to assess accounts, understand their payment status, and determine best potential actions.
In this example, the manager can see that overall outstanding receivables equal US$85,807 (lower left) and of that amount, $33,377, or 38.90% of all outstanding receivables, is overdue (upper left).
Two metrics concerning “overdue receivables” are provided to help provide context:
- Top overdue customers: A ranking of the top five customers by their overdue balance amount.
- Overdue by days: A grouping of overdue amounts according to the number of days overdue (1-15 days, 16-30 days, and so on).
Next, two metrics highlighting “future due in receivables” provide information on invoices that are open but not yet due:
- Incoming by days: A grouping of all pending invoices according to when they are due (1-15 days, 16-30 days, and so on)
- Payment predictions: A prediction of the number of days it will take for any invoice to be paid
Finally, the table at the bottom lists specific invoices. Drilling down, the AR manager can directly access invoices in the underlying finance or ERP system to understand payment terms, review the products involved, or look up customer contact information for appropriate follow up on late payments.
Optimizing collections with augmented analytics
Based on the metrics above, the collections manager can gain insights and take appropriate action. From below, AR has a clear picture of payment predictions and can see that $2,478 will most likely be tied up with customers who will be at least 90 days overdue.
If optimizing collections is the goal, this a good place to start. Fortunately, information presented at the aggregate level can also be viewed by individual customer so AR can identify customers in this category and determine the best strategy for the business – for example, set up a slightly more aggressive collection approach or consult with the sales team regarding the customer relationship and enlist their help in taking further action.
Late payment predictions are based on a machine learning model that gets more accurate with more data, so as the collections manager feeds results data back into the model, it gets smarter. For example, did the adjusted collections strategy work with the customer in question? Did the situation improve (or get worse) for some other reason? Whatever the case, the machine learning model can take this data into consideration to make better predictions in the future – learning from the actions and data – and ultimately helping to drive a more effective collections strategy.
Gaining insights to see what’s ahead
It’s important to remember that the predictive model illustrated here is not merely a mechanism for identifying problem customers. The model can also predict late payments for customers in good standing who have always paid on time. It all depends on the richness of the data sets incorporated into the model.
For example, if a key customer is linked to a supply chain partner based in a region recently hit by an earthquake, a currency devaluation, or an epidemic like the coronavirus, an effective machine learning model with the right data set could bring such factors into its calculations and alert the collections manager to potential risks. Then, corrective action can be taken even before the company feels the impact. That’s the power of augmented analytics.