Peugeot Tailors Content in Real Time to Attract Consumers to its Showroom
PSA Peugeot Citroën, the automobile company, was struggling to drive consumers to its showroom. Ranked ninth globally in terms of production, the company had recently taken a billion-Euro bailout, and was in the process of plotting a return to the U.S. market after nearly 20 years. The company was marketing aggressively in Europe, its primary market, but seeing little in terms of increased foot traffic at dealerships.
One executive spoke about her results with the media saying, “We have multiple agencies. Every time we meet with them, their recommendation is to spend more money. So we raise our budget. But the only ones making any money are the agencies themselves. We just aren’t selling enough cars.”
This is a common refrain among marketers, and was the main concern for other top executives of PSA Groupe. It is natural to equate increased marketing spend with improved sales results, but in a world where consumers are hard to identify and convince, Peugeot knew it would take more than increased spending to reignite conversation and consideration for the brand.
Samir El Hammami, manager of Peugeot Citroën’s digital marketing team, had an abundance of website data on consumers who came to PSA sites to browse inventory and use the custom car configuration tool to build and price auto models — but consumers visiting the site often left before they were fully
engaged. Site “dwell times” were down, and most consumers left the website without requesting more information.
For most auto retailers, an in-person dealership visit was the ultimate signal of intent; drivers that Peugeot could persuade to move from site visitor to test driver were much more likely to purchase. How could Samir use real-time people data to personalize the consumer’s experience on mobile and web properties?
Samir thought a more real-time application of people data could create stickier web experiences with
the brands and result in more test drives. “My goal was to capture and analyze every interaction on the site — by car type, color, specification, and price — and use it to improve the number of consumers
signing up for a test drive, which was my most critical performance metric,” said Samir. “It hadn’t really been done before, so it would require a big leap.”
The challenge was connecting PSA’s unique and granular customer segmentation with the most relevant content, and delivering those experiences in the moment through the car company’s customer content management system. This effort focused on the Peugeot, Citroën and DS brands. We analyzed user behavior, activities, and events as they occurred in the users’ browsers and computed microsegments and user affinities for various attributes — make, model, color, type, etc. — in real time.
How could we compute all this information in a world where milliseconds mattered and browser-server round trips were too time-consuming? We met the challenge by putting the core affinity computation module into the browser and caching and updating it periodically, according to the frequency with which the user engaged with PSA’s content.
In essence, we enabled PSA to give the user on the other side of the screen an experience tuned perfectly to their interests. A data-mining technique, frequent pattern analysis, identified patterns from a large data set. Frequent pattern analysis is a great complement to A/B testing, a method many marketers already use, to compare the effectiveness of certain web pages or advertisements in a controlled experimental design. Here, frequent pattern analysis gave Peugeot the ability to test multiple hypotheses at once and reduce the amount of creative testing required so that the company could give its users a narrower, more promising array of options.