Wouldn’t it be great if we could accurately identify which customers or prospects are going to default? We would be able to refuse or restrict credit terms and prevent the businesses we represented from being stung by insolvency; we would be nothing short of psychic credit superheroes! But is it possible?
Unless you’ve been living under a rock, you’ll have heard of the rise of predictive analytics. If you’re already using credit reports and relying on credit scores, you’ll be using predictive analytics to some extent.
Predictive analytics are based on algorithms. Algorithms determine the patterns in data and their relative importance in establishing a decision or outcome. In the context of credit reports, an algorithm will determine the credit score assigned to a business or individual and the integrity of that scoring will be based on the quality of the algorithm.
This is the reason why we might have differing views about the quality of credit reports we subscribe to. It all depends on the algorithm; how comprehensive it is and how relevant it is to our business.
Clearly, using credit reports as part of a diligence process is important, but can a business improve upon the predictive integrity by adopting its own approach to predictive analytics. Can it outperform the credit reference agencies?
The answer is yes, at least in theory. Each business will have its own idiosyncrasies in its customer base. A business might have a bias towards customers in a particular geographic region or industry. It may tend to serve businesses of a certain age or size. This is important because it has an impact on the relevance of algorithms. For example, some credit reference agencies will lower a credit score when directors resign from a company. Whilst this might be very relevant for SME’s (where the loss of a director can be critical), it is less relevant for large enterprises. As a consequence, if you’re doing business exclusively with large enterprises, director resignations may have little or no significance to your credit risk.
Also, whilst credit reference agency data is undoubtedly important, so is the wealth of data you hold (although only if you’re actually making use of it!).
As any episode of CSI will tell you, one of the biggest predictors of the future is relevant past behaviour. If you have years of payment data in your organisation, you should be able to use that to predict who will pay late, who needs chasing (and to what extent) and even insolvency events.
So what are typical ‘insightful’ indicators?. Here are just a few:
- Number of late payments in the last 90, 120 and 360 days
- Payment method and changes in method
- Industry/SIC Code
- Size of invoice
- Current balance vs credit limit
There can be other, less obvious indicators; producing patterns that only data-mining software can identify.
Using Salesforce Einstein Prediction Builder, users can very quickly build data models that can assign a probability score for late payment and critical default. Once data is housed on the platform, building predictions is simple. Prediction indicator fields can then be used to produce alerts or help prioritise accounts for credit control calls or credit limit reviews.
Einstein Prediction Builder forms part of a much larger use case for Salesforce, which includes automating dunning cycles, allowing customers to view and manage online accounts and live KPI dashboards.
For more information on predictive analytics for credit managers, contact firstname.lastname@example.org.