The trend toward smaller mobile devices has changed channel dynamics for Intel. In the past, companies bought parts from Intel and built their own computers. As a result, "we had a very clear view of the whole market," says Lilian Kubail, sales insights manager within Intel's sales enablement organization. Now large OEMs are building smaller laptops and tablets and then selling those products to distributors, who in turn sell to resellers, making it harder for Intel to track who is selling its products.
To help sales staffers find those accounts and engage with them, Intel developed its own advanced predictive analytics solution.
The project involved combining existing information on resellers with external data on which organizations are making big investments in the market and growing their businesses. Then the IT team looked at interactions with Intel's websites to see which organizations were looking for information, and which products were being accessed.
Once data was gathered, "the tricky part was figuring out which data sets were the most relevant," says Ivan Harrow, director of business analytics in sales and marketing IT.
Peleg Havazelet
Intel's big data team: Seated: Jeremie Dreyfuss, Ronen Aharony and Dafna Yaari.
Standing: Asaf Cohen, Ido Bar-on, Maya Shamir, Eiran Bolless, Shani Chaushu, Einat Ronen and Itamar Snir.
The analytics engine takes the data and uses a methodology called "unsupervised clustering," which looks for hidden patterns in unlabeled data. Once the clusters have been defined, it goes through "supervised classification" where business rules are applied to the data, and then the output is ranked from highest to lowest sales volume potential.
The predictive analytics engine was tested at Intel's Asia-Pacific online sales center in late 2012 and then rolled out globally. It yielded an annual increase in incremental revenue of $20 million.