In today’s economy, marketers are under pressure to do more with smaller budgets and demonstrate contribution to sales and profits. More so than ever before, they need to prove the effectiveness of their campaigns by measuring results in order to justify their marketing spend. How can this be done? Predictive analytics is the art and science of building statistical models that predict customer behaviour. By harnessing predictive analytics, automotive marketers can turn customer data into actionable insights that increase campaign profitability and improve the return on marketing investment.

Specifically, predictive analytics can provide answers to four important questions:

1. Which customers should I target?
2. What products should I offer the customer?
3. How much of an incentive, if any, do I need to offer for that customer to convert?
4. When should I make the offer to the customer?

Predictive models target the right customers

Predictive models are the output of the predictive analytics process. These models allow marketers to target the right customers, weeding out the customers who are not likely to respond, identifying what products or services a customer is in market for (e.g., a new vehicle, a used vehicle, service, etc.), predicting what specific products they might be interested in (e.g., vehicle segment or model-level preferences), or determining whether and how much of an incremental incentive is required to entice the customer to convert.

Models can be enhanced by augmenting the customer data a manufacturer has access to within their own databases with demographic and lifestyle data from third parties. When doing so, automotive marketers have to be mindful of the data protection legislation. Where this legislation is restrictive - Italy, Spain, Germany - it limits the availability, quality and use of demographic appends. While good models can be built without third-party data, access to it simply enhances the quality and accuracy of the model.

In the example below, a “propensity to buy a new vehicle” model built with only data the automotive manufacturer had in their database allowed us to predict 79.5% of sales in the first half the model. By incorporating demographic and lifestyle data from third parties, the accuracy of that model improved to 86% of sales (see Figure 1).

Know when to make an offer, what product to promote and what incentive to offer

The following real-life examples show how predictive analytics can help make more informed, and better performing, campaigns for both new vehicles and service.

Pumping Up New Vehicle Sales

One client wished to launch a direct mail campaign to drive sales for a new vehicle product launch. The budget would support a maximum volume of 200,000 customers. Prior campaign results were not available, so we built both an In-Market Timing model and Product Affinity model to hone in on customers most likely to respond to that campaign. We scored the entire customer base on their propensity to be in the market for a new car in the next six months, as well as their affinity towards purchasing the car model that was being offered. By using these two models in combination, the client was able to target the most appropriate customers and achieve the highest ROI on that campaign (see Figure 2).

Driving More Customers to the Service Bay

Many manufacturers run recurring service marketing campaigns, so the results of previous campaigns can be used to target the most promising customers. For one client, we built two models - a Propensity-to-Respond model and an Estimated Spend model.

The Propensity-to-Respond model scored each customer based on their likelihood to respond to the offer while the Estimated Spend model determined how much money the customer was likely to spend on his next service. By overlaying these models we were not only able to identify the most promising customers, but also identify which incremental offers would still provide a positive ROI.

The client initially planned to offer a tiered discount, with customers in the lower spend segments (S1 and S2) being offered half the discount amount of those in the higher spend segments (S3 and S4). This offer would have resulted in a loss for customer segments S1 and S3 (see Figure 3).

As a result of this analysis, we recommended a different offer structure: free inspection to S1 customers and a tiered discount where S2 and S3 customers received half the discount amount offered to S4 customers. This offer structure ensured that all the customers returned a profit and the profitability of the campaign as a whole was maximised, as shown in Figure 4.

Continuous Improvement

Campaign results should be fed back into the predictive models to continually refine their accuracy, and hence the ROI on future marketing efforts. By leveraging predictive analytics, OEMs can better plan campaigns that will drive service sales and ROI and build a base of engaged customers to fuel future vehicle sales. The ability to not only measure campaign effectiveness, but to continually improve it will allow marketers to persuade the CFO/CEO to increase marketing spend versus cutting it.

Click here to contact us and learn more from an Urban Science expert