Use Your Data to Predict What Customers Will Do

“Predictive analytics” sounds scientific and, well, a little scary.

But as what eventually happens with new technology, predictive analytics is being democratized with new SaaS (software as a Service) offerings. They mask underlying complexities and make analytic models accessible to virtually anyone.

This is good news for automotive brands, agencies and dealerships that now have easy access to a host of previously inaccessible sophisticated marketing techniques.

The primary driver for use of predictive analytics is consumer expectations. A few years ago, an SUV shopper who received an email promotion for a minivan would just delete the message.

Today, that same prospect will still delete the email but then potentially post something on Facebook about how your dealership wasted his or her time with an irrelevant email. That miffed person may decide to buy an SUV elsewhere. 

“We live in a world of 7 billion ‘me’s,’” Forrester analyst Mike Gualtieri writes in a report titled “Predictive Apps Are The Next Big Thing In Customer Engagement.”

“Customers increasingly expect and deserve to have a personal relationship with the hundreds of firms in their lives,” he says. “Companies that continuously ratchet up individualization will succeed. Those that don’t will increasingly become strangers to their customers.”

The advent of online shopping and buying has enabled companies to easily collect behavioral data, which has proven to be a much better predictor of future buying behavior than segment profiles alone.

Examples of behavioral data include whether a person clicked on a link, visited certain web pages or requested information on a certain vehicle.

Predictive analytics enables marketers to make informed decisions based on both behavioral and segment data.

It involves creating models that leverage existing data sources to predict what customers will do next. For example, what vehicles are customers most likely to buy, and what is their likely purchasing timeframe?

Consider a dealership that wants to determine which customers may purchase in the next 90 days.

The store might evaluate the following customer behavior and profile data: number of sales over five years; amount of repair orders over two years; and clicks, openings and browsing behavior over 18 months.

The model crunches the data and assigns a value to each customer based upon how their actions correlate to the actions of prior buyers.

The score essentially is the probability of something happening, scaled in a predefined way. In this example, highest-scoring customers have the greatest probability of purchasing a car within the next 90 days, and hence are good prospects for a campaign.

A dealer can get started with predictive analytics by buying packaged models. Most dealerships have some sort of email marketing solution in place. Those already allow for some level of personalization. Robust ecosystems of add-ons have emerged around the most popular email marketing platforms.

If you already have some sort of customer engagement or marketing platform in-house, ask your vendor if they offer packaged analytics models. Generic models exist, and some vendors are now offering models developed specifically for the automotive industry.

The most common used in the automotive industry are designed to predict the likelihood of a customer:

Imagine how much more targeted and cost-effective campaigns would be with this information in hand. An Aberdeen Group study found that campaigns based on predictive analytics resulted in an 8.3% incremental sales lift.

This isn’t about “big data,” but rather actionable data that helps drive incremental sales. Predictive analytics is transforming the way automotive brands and marketers engage with customers, and it’s now within dealers’ reach.

Mike Wethington is CEO of Minneapolis-based Outsell.