Use Customer payment predictions

This article explains how to use the Customer payment predictions. Before you use this feature, make sure that you've completed the setup steps for it. For more information, see Enable customer payment predictions.

You can view customer payment predictions in the Manage customer credit and collections workspace and on two new list pages: Transaction payment predictions and Customer payment predictions.

Manage customer credit and collections workspace

The Manage customer credit and collections workspace includes two new tiles: Transaction payment predictions and Customer payment predictions.

Transaction payment predictions list page

On the Transaction payment predictions list page, you can view the probability of payment for open transactions in the On time, Late, and Very late buckets. For each transaction in the grid, the On time probability column shows the probability that the invoice will be paid on or before the due date. If the probability of an on-time payment is less than 50 percent, a red circle appears next to the percentage in the On time probability column to indicate the risk of late payment.

The Related information pane on the right side of the page shows more details about the predictions:

• For the transaction that is selected in the grid, the Payment prediction FastTab shows the details of the payment predictions in the On time, Late, and Very late buckets. The Top factors section shows the top factors that influenced the predictions. The top factors are attributes of the selected transaction and/or the customer for that transaction.
• The Customer insights FastTab shows the current invoice, payment, and collections statistics for the customer for the selected transaction.
• The Customer history FastTab shows the customer's payment history in the On time, Late, and Very late buckets.

The data in the Top factors section, and on the Customer insights and Customer history FastTabs, helps explain the payment predictions. It can help increase your confidence in the efficacy of the predictions.

Customer payment predictions list page

The Customer payment predictions list page shows the total open balance, and the amount that is predicted to be paid in the On time, Late, and Very late buckets.

The payment amount in each bucket is calculated as the sum of the weighted average of the transaction balance. This amount is calculated based on the payment probabilities in each bucket.

For example, a customer has three open transactions that have the following payment probabilities in each bucket.

Transaction	Amount	On-time payment probability	Late payment probability	Very late payment probability
T1	100	10 percent	50 percent	40 percent
T2	1,000	50 percent	30 percent	20 percent
T3	10,000	1 percent	4 percent	95 percent

In this case, payments are projected for each bucket in the following way.

Buckets	Transaction T1	Transaction T2	Transaction T3	Total
On time	100 × 10 ÷ 100 = 10	1,000 × 50 ÷ 100 = 500	10,000 × 1 ÷ 100 = 100	610
Late	100 × 50 ÷ 100 = 50	1,000 × 30 ÷ 100 = 300	10,000 × 4 ÷ 100 = 400	750
Very late	100 × 40 ÷ 100 = 40	1,000 × 20 ÷ 100 = 200	10,000 × 95 ÷ 100 = 9,500	9,740

The Related information section on the right side of the page shows more details about the predictions:

• For the transaction that is selected in the grid, the Payment predictions FastTab shows the details of the payment predictions in the On time, Late, and Very Late buckets.
• The Customer insights FastTab shows the current invoice, payment, and collections statistics for the customer for the selected transaction.
• The Customer history FastTab shows the customer's payment history in the On time, Late, and Very late buckets.

The data on the Customer insights and Customer history FastTabs helps explain the payment predictions. It can help increase your confidence in the efficacy of the predictions.

Improving the accuracy of payment predictions

You can view the accuracy of payment predictions by going to Credit and collections > Setup > Finance insights > Finance insights parameters. On the Customer payment insights tab, the Prediction model section shows the accuracy of the prediction model as a percentage.

If you aren't satisfied with the accuracy, select the Improve model accuracy link to open the AI Builder extension experience. In the AI Builder extension experience, you can select or cancel the selection of fields until you've selected the fields that you believe are most important for accurately predicting payment probabilities. When you've finished, you can easily retrain the prediction model and publish your changes. The newly trained prediction model will automatically be picked up for predictions in Dynamics 365 Finance.