Predicting Your Next Car Purchase
By Joe La Feir, SVP, IS&S (Information Systems & Solutions), IHS Automotive
Joe La Feir, SVP, IS&S (Information Systems & Solutions), IHS Automotive
Last year I recall a friend of mine saying it's crazy that it seems like every other house in his neighborhood had a Chevrolet Corvette in their garage. My friend is a 45-year-old professional living in suburban Detroit. He is married with two teenagers. During the last 20 years, he and his wife have seen 19 vehicles come and go in their own garage. He is fairly particular about his vehicle choices and has always had domestic or German brands. Not surprisingly, in 2000 after their first child was born, they leased three minivans in a row.
“Big Data is changing the way the automotive industry does business. OEMs, their marketing teams and agencies, are using Big Data to their advantage in the sales process”
Recently he sold a Porsche that he had owned for about five years. Six months later, he pulled into his driveway with a sleek new Corvette Stingray. He told me that he had spent significant time researching his new ride online, using configuration websites and even the Chevrolet website for almost a year ahead of the purchase. The funny thing about his experience is that he does not remember ever getting targeted through a marketing campaign to purchase a new Stingray.
At the same time, he was getting ambushed with some really impressive campaigns from Porsche, including hardcover books with beautiful pictures of the latest Porsches. He also received a really cool USB memory stick in the shape of the car. It is clear that these must have been some expensive marketing campaigns; yet he heard nothing from Chevrolet, which is surprising.
In the case of the Corvette Stingray, the product did all the work for General Motors. However, most of the time, the consumer has a much more complicated decision to make regarding their next vehicle purchase. The fact of the matter is that marketers spend a disproportionate amount of their marketing dollars retaining existing customers as the return on the investment is so much greater; it costs significantly more to conquest a new customer than to retain an existing customer.
Retaining customers is less expensive in part because it is managed from internal systems at the OEM, namely CRM, and these customers are known. Conquest efforts are traditionally sporadic, sourced from varied audience/ marketing providers, and measured inconsistently. Lack of systemization around conquest efforts is one key reason it is much more expensive. According to an IHS Automotive analysis of vehicle registrations, customer conquests represent 50 percent
Why is it so hard to target these customers? There are several challenges facing automobile companies when they try to target conquests. The virtual sea of prospects, little information about what is motivating a customer to switch, and their decision on which of the hundreds of vehicles they will choose are all competing factors. The majority of these consumers are unfamiliar to a brand, and understanding these key questions would allow conquest campaigns to be more targeted and focused on the highest quality prospects during their peak shopping time.
This is where Big Data and advanced analytics come in. My friend’s experience only scratches the surface on the number of data points that are available to model consumer behavior. Today, there are literally billions of data points representing information on well over 120 million households in the U.S. Some of these include: online shopping, customer survey, pricing, incentive and recall data.
Big Data is changing the way the automotive industry does business. OEMs, their marketing teams and agencies, are taking a closer look at Big Data and using it to their advantage in the sales process.
One example that IHS Automotive customers bring to us regularly is the question of what is motivating a customer to switch. Using Big Data technologies to bring together many large sources of data, we can now begin to understand the rationale behind that motivation. By connecting data points that highlight genuine customer feedback and through surveys and transactional data, we can learn a significant amount of detail. For instance, if surveys reveal that Vehicle X has a large number of customers stating dissatisfaction with handling and cargo space but yet are extremely satisfied with the exterior styling, are these drivers of loyalty or defection? Not necessarily. With this data alone, we simply cannot tell.
However, if we were to connect this data with the next purchase, and we see that a high percentage of Vehicle X owners are switching to Vehicle Y when they rate handling low, and we see high rating for handling on Vehicle Y, we now know that it is a key driver of defection.
Conversely, we may also see that defection is not higher when cargo space is rated low. While it is not a feature of the vehicle that is liked in this example, it is also not a driver of defection. In the example, we also can see that marketers for Vehicle Y may want to target consumers who drive Vehicle X as conquest targets and perhaps with specific messaging. Also, from a product perspective, it’s easy to see how this information provides valuable insight into prioritization of product changes that will have the biggest impact.
Looking at the other challenges, Big Data technologies are allowing automotive marketers to analyze decades of vehicle purchase history to help determine when conquest targets will be in market and what they are likely to buy. Connecting and analyzing longitudinal data for decades of vehicle purchases in every household reveals patterns of repurchase timing and segment changes. My friend’s history of 19 vehicles over 20 years, three minivans in a row, and the recent transition to a new sports car, is very powerful data in predicting what he and others like him are likely to do next. Taking this to the next level —as such recent online vehicle shopping activity for the household and others that “look” like this household—we are able to unlock tremendous predictive power.
This capability does not come without complexities. First and foremost, when dealing with such large volumes of consumer data, privacy must be considered and built into the system from the start. Techniques such as anonymizing the data with linking technology to remove personal identifiable information in the Big Data platform and the use of aggregates can be good methods for dealing with this issue.
Secondarily, the Big Data platform should not be treated as a dumping ground for data to be used in the future. Time should be spent up front considering data linking and data aggregation strategies. At IHS Automotive, we actively pursue these opportunities. A carefully orchestrated Big Data platform will provide a powerful data environment for data scientists who conduct research and modeling, and will ease the introduction of new resources as they become available.
With Big Data and advanced analytics, we are able knit together billions of data points to provide incredible predictive power. This unprecedented level of knowledge helps the automotive industry get the right message to the right consumers, ultimately driving key vehicle purchase decisions that positively impact the bottom line.
