Artificial Intelligence and Machine Learning (AI/ML) are current buzzwords and many companies are now investigating the possibilities across various business functions. These technologies can have significant impact upon the performance of credit management but it is also easy for companies to get it wrong. What I want to do with this article is briefly discuss the key impact areas and highlight some issues that credit managers need to be aware of if they are going to get it right.

Collection processes

Perhaps the best-known software type on the market is collections software. There are over 100 collections software vendors and their technology, functionality, investment levels, service levels and pricing models all vary. How are you to know which vendors to shortlist? You probably won’t know enough about technology and your IT and procurement teams won’t have a detailed understanding of credit management and its potential contribution. We have seen a number of companies which have run RFP processes in which the initial shortlist comprised inappropriate vendors. From that point on, a less-than-optimum outcome is
designed into the whole process.

One of the first issues to consider is that you may not even need AI/ML. You may simply want to improve efficiency (or that may be all that your CFO wants to see!). Considerations such as customer experience may be less important. In this case you should be looking for something “cheap and cheerful” and which “does what it says on the tin”. There are some good-quality solutions that deliver exactly that. The best solutions use AI/ML to analyse customer behaviours in order to determine the most effective communication methods, style and timing to maximise customer response. They enable your collectors to target their work to where they add the greatest value in the customer interface. And they motivate your collectors by giving them the tools to do the job professionally – this can be a key issue in shared service centres to help reduce costly staff turnover.

Cash allocation

Cash allocation software can have a big impact. The very best solutions actively use machine learning to achieve exceptionally high auto-matching rates. Normally the business case will be simple, based upon a headcount savings calculation. But if it is important to you, it can also fit within a “joined-up” approach.

For example, you don’t want to be attempting to collect from customers who have already paid you. And you may also want to pick up additional risk indicators such as the customer using a different bank account to pay you.


The current risk outlook (and reduction in credit insurer limits) has made this a topical issue for credit managers. We are seeing that companies want to review or “beef up” their credit risk management processes. This can sometimes be to ensure compliance (internal and external such as credit insurance) but it can also be because there is a desire to drive profitable (or speedy) sales on the back of finely tuned risk assessment. At Co-pilot we talk about “The Three Pillars of Credit Risk Management”:

• Financials and analysis
• Big data (internal and external)
• Market intelligence

This requires the collation of huge amounts of data but it has become much quicker and cheaper to do so. Doing this and automating decisioning and review processes for low-value high volume accounts enables efficient sales/risk strategies to be set and allow your credit analysts to focus on the key risks where they will want best-available intelligence and analysis. Workflows can be designed to give them red flags (for example late filing of accounts). Many software vendors claim credit risk management functionality but often it only goes as far as supporting risk-weighted collections strategies.

What we would regard as “true” risk functionality is delivered by vendors who truly understand risk and also the importance of some key principles, for example: how the platform interfaces with other service providers used by the credit manager. There are only a small number of credit risk management software vendors who pass the “Co-pilot test”.

Customer behaviour analytics tools – drive retention and sales

Big data enables immensely detailed analysis of customer behaviours. This type of solution works particularly well for companies which have large ledgers and multiple brands and product lines. You can identify customers who are likely to leave you, enabling your company to target its retention campaigns. It will also identify customers to whom you can cross-sell or upsell. Imagine the impact if this is linked to credit and fraud risk management technology that enables the business to ensure that credit limits are in place before the sale. These behaviour analytics tools are primarily targeted at sales and marketing directors but credit managers may feel that they have a legitimate interest in bringing them to the board’s attention as part of a “joined-up” business strategy. It is this kind of behaviour that means the credit manager is visibly operating at director level.


So, what is the key take away here? Well it is that the first steps are the most important ones and it is better if the credit manager is initiating the process. This requires a bit of work, but it is a healthy process. You need to be sure that you are doing the right thing for the right reasons and that you have a strong business case which supports your company’s key objectives. Preparing the business case is often the opportunity to think through and then tangibly demonstrate the credit management function’s impact upon efficiency, sales, cash position, customer experience (reputation) and Return on Investment. Once you are clear about what you want to do, you need to bring on board key stakeholders such as IT and procurement. But you can do it in a way that they are reacting to your well-articulated ideas, not the other way round.

Simon Marshall FCICM is Managing Director of Co-pilot UK.