Collecting and selling data about people is estimated to be a $200 billion business, and all signs point to continued growth of the data-brokerage business.
Here’s how the business works when the data is being sold to consumer marketers. Data brokers collect information about customers wherever they can: through loyalty cards, public records, social media posts, and most often by tracking their browsing behavior across different websites. All this tracked customer information is then fed to machine-learning algorithms, which build segmented profiles of similar groups of people. These digital profiles are then packaged as “audiences”; typical groupings might be “fashion interested” or “males 25-54.” Marketers can buy these off-the-shelf audiences from data brokers for ad targeting. For example, Nike or Adidas might buy access to the audience segment “fashion interested” to communicate with prospective sneakers buyers.
But there’s a problem. The process by which data brokers create these segments is kept secret for competitive reasons. This unfortunately means that marketing managers do not know whether they can trust the audience data, despite the fact that these purchases typically make up a large portion of their media budgets. New York Times CEO Mark Thompson once famously asked: “When we say a member of the audience is a female fashionista aged 20 to 30, what’s the probability that that’s actually true?”
We recently tested the accuracy of popular audience segments that a range of brokers have on offer. We looked at age and gender, as well as customer interests such as sports interested, travel interested, and fitness interested. To check accuracy, we used data from settings where people had revealed this information voluntarily. For example, we compared the segment characteristics with consumer data provided by survey panels, which was also cross-validated with information from Facebook or from a financial institution.
Across all our tests, we found…