Keeping data in-house rather than sending it to the cloud has advantages, says Marc Gallman, senior manager, big data architecture at Lenovo. One advantage is creating a nontraditional attribution data model that continually proves a positive return on investment.
Lenovo scaled back its attribution modeling and the data it sees to make campaigns run more efficiently. Rather than view all the data, company marketers focus on Web analytics data without including all the peripheral data sets such as who clicks on the email and interacts with it. "The consumer might see 10 ads before coming to the site, but we don't see that data," said Gallman.
"Some attribution models include the impression data outside of the Web property, but we're not seeing a ton a value by including that," Gallman said.
Understanding how to model the data for attribution has become one of Lenovo's biggest challenges, especially because the company brings in data from search, email, display advertising, and more than a dozen other sources.
advertisement
advertisement
"When you start asking questions and get into the granularity of the data, you start to bring in very large data sets," Gallman said. "An ad impression data set is easily in the trillions for many companies."
Lenovo also went against the grain by not moving all its data to the cloud. Working with Talend, a big data integration company, allows Lenovo to bring in and process data from a variety of sources in house globally. "I like that it's not a 100% cloud-based model," Gallman said. "I wanted it to execute on our machines and I didn't want our data to leave our infrastructure."
Bringing the data in house created another challenge to turn the company's Web analytics team into a big data team since it began using technology from Talend to analyze more than 22 billion transactions and 11 billion customer records annually, monitoring purchasing behaviors and the buyer’s journey across multiple channels.
Lenovo brings in about 250 terabytes from more than 15 worldwide sources such as advertising, call center data, marketing data, CRM, social, surveys, and transactions. The process has resulted in a savings of $1 million in operating costs in the first year and an 11% average revenue per unit increase.
Prior to using Talend, everything was hand-coded, Gallman said, so it's not difficult to understand how Lenovo saw a positive return on its investment in between five and six months.
Lenovo has been working with Talend for just over three years, and the biggest challenge has been to find people who know how to use the platform.