Predictive analytics is the crystal ball of the corporate world: you can often predict the chance that a customer will become insolvent or decide to halt business operations and it helps you make decisions about doing business with a promising start-up. This crystal ball is essentially a gold mine for you as a credit manager or marketeer. But how and when do you deploy predictive analytics?
Predictive analytics begins with collecting data, where you have to ensure an effective data management policy. This data can be derived from your own company, which already has a lot of customer information about loyalty, the number of contracts, revenue and visits among other things. The drawback here is that your stored data is often subjective or incomplete. It’s therefore advisable to supplement internal information with data from external sources and to have your database examined. Think of supplementing your customer database with demographic characteristics and credit scores, the profile of your best customer and opportunities for cross-selling and up-selling. The more data you can collect, the more accurate your predictions will be.
Do you have sufficient data to hand? If so, look for patterns or correlations. But don’t be too intuitive. Take the following example:
You see that there’s a connection between orders placed by a customer and the number of times they have been visited by their account manager. You can jump to premature conclusions on the basis of this one correlation alone. This is where predictive analytics come into play: these help you find patterns and connections which aren’t immediately observable.
In this way you can predict meticulously how big the odds are of your new business prospects actually becoming customers. You’ll see that your customers are similar. So compare your ‘average’ customer to your prospect. The more characteristics they share, the bigger the odds of your lead becoming a customer.
Walmart is a good practical example of predictive analytics. They discovered that consumers stock up on gingerbread, spare batteries and torches when bad weather is approaching. This information allowed Walmart to generate more sales. As a result of their predictive analytics, Walmart stocked up on torches during bad weather, and managed their stock more efficiently.
The example shows that collecting data can provide surprising insights. But your company can only benefit from this if you respond quickly to the predictions - even if they’re negative. This allows you to immediately take action to prevent possible damage. Predictive analytics can help you to anticipate future events.
It’s good to know that the more frequently you carry out predictive analytics, the more reliable the results become. You can also check whether the predictions were actually correct by adding the results to your existing database. This makes your system even more intelligent and your predictions even more precise.
Predictive analytics also play a big role at Graydon. An example of this is the Graydon rating, which predicts the likelihood of default. This chance is calculated on the basis of advanced statistical models which examine financial statements, payment behaviour, demographical characteristics of the company (such as the size, age and industry) and exceptional situations within a company. This results in the Graydon rating, which expresses the chance of insolvency within 1 year. The rating is then subdivided into one of the 11 possible ratings, which are: AAA, AA, A, BBB, BB, B, CCC, CC, C, D, NR. AAA indicates the lowest credit risk, D indicates that the company is insolvent and NR indicates that no rating has been determined.