As a marketeer, you are always looking for tools to gain more insight into your customers and their behaviour through data. Predictive modelling began its rise a number of years ago and has since become a reliable tool for predicting customer behaviour and needs in order to adapt your business operations, products, services and advertising accordingly. Three applications for marketeers.

Predictive modelling

Simply put, predictive modelling is a prediction model, a calculation of algorithms with which you try to predict the truth as well as possible. It is a model, not a one-to-one truth. Use it as a tool to gain insight.

In order to build a good prediction model, you can use both data from your own organisation and data you purchase. Important to know as a marketeer when you start working with predictive modelling: in most cases, it is better to look at the behaviour of your consumers than to divide them into general consumer profiles. If you do this, then base the profiles on your own customer base. The clicking and buying behaviour of your own customers says more about follow-up actions than a general label of who your target audience is. Three applications of predictive modelling for marketeers:


1. Assortment analysis

Predictive modelling can be applied to predict what products or services your customers purchase and what they may purchase in addition. You can use your own historical ordering or purchasing behaviour for this, analyse it and build a prediction model based on that. Some examples. A sports brand can predict what models and colours they need to make based on purchased items. An online web shop can predict what accessories and clothing items female consumers want to purchase with a clothing item and directly display it as a suggestion below. The possibilities are endless.


2. Churn rate

Predicting how great the chance is that people terminate the services with your company as a customer or member is also very possible with predictive modelling. From the moment a consumer becomes your customer, you can analyse their behaviour. Based on that analysis, you can predict when someone is likely to leave. You can also link your data to information about houses being for sale. This can be relevant to telecom companies that offer cable or TV packages. You can also make analyses and predictions based on mailing behaviour. If you have insight into when someone is considering leaving, you can initiate loyalty programmes, for instance.


3. Customer acceptance

Predictive modelling can also offer a solution in the customer acceptance process. Start with an analysis of the (payment) behaviour of the customers from your own file. Filter the customers based on well and not so well paying customers and see what properties they have. Where do you see patterns? You can then also purchase data to help you make valuable analyses and predictions. Think of linking postal code data to payment behaviour. That knowledge can be automated and used to predict which customers have an increased payment risk. This allows you to set up additional conditions for new customers in the acceptance procedure.