Blended Data: A Big Player in…

Blended Data: A Big Player in the Future of Profile-Building

Top Take-Aways from Programmatic I/O


Big data is so big that IBM estimated that we collectively generate 2.5 quintillion bytes of data on a daily basis — and that was two years ago. Technology has since evolved, allowing us to collect and manage data more efficiently. It only makes sense that the way we actually use that data should follow suit.

During last week’s Programmatic I/O conference, members of the “The Next Generation of Profile-Building Data” panel discussed various key elements of data success, including real-time data. Kathy Leake, CEO of Qualia, points out how stale data can become after not being analyzed for 30, 60, 90 days. Another touchstone, she adds, is combining the data; this creates a more accurate picture of customers, rather than pigeonholing them because of an isolated action.

We should blend data, versus binary decisioning, because most of ad tech is based on a consumer exhibiting one behavior one time on one device,” says Leake. “Segmenting will be better once we actually weight our signals and score our signals.”

That kind of one-time segmenting isn’t cost-effective. One of the most common examples for marketers deploying data is the pair of shoes you looked at that proceeds to follow you all over the Internet. But what if you only clicked the link because your friend said, “Check out the shoes I just bought!” but you actually think they’re ugly? The brand is paying to repeatedly show you a product that you’ll never buy.

On the other hand, combining data leads to better-targeted ads, something that ultimately benefits the consumer.

“The Internet is free because of advertising,” says Joe Zito, vice president and general manager of retail at Oracle Data Cloud. “It’s not going away; do you want it to be relevant, so I’m no longer seeing women’s yoga pants from Lululemon. Make [ads] useful and relevant to me as the consumer, and to the marketers so they’re spending their dollars the best, and all places in between.”

That practice falls in line with the learning curve that accompanies all new technology. It’s not that current data management platforms (DMPs) aren’t capturing the best data. Most marketers just aren’t using them as effectively as they could, as a result of not having caught up yet.

One of the most important things for a marketer to learn about a DMP is when not to rely on the technology. Though programmatic is obviously an invaluable component of data-driven marketing, Michael Horn, chief analytics officer at Resonate, recommends a mix of machine analysis and human judgment.

“For marketers, many of whom have a challenge attracting talent, they don’t necessarily have a lot of programmatic natives to answer human readable questions and get human insights,” he says. “A lot of the challenge is making sure companies have unique data and are able to contribute those to the DMP and provide an interpretive layer to sit on top of them for when the DMP can’t understand the strategic questions.”

For example, “social listening” can provide insights on buyer behavior, but it can also require human interpretation. Lots of people order pizza on family movie night, so a brand like Domino’s may be interested in what people are posting about on Friday evening. But while machines can monitor Facebook and Twitter feeds for certain keywords, they can’t extrapolate in the same manner as a person.

No matter what tactics marketers employ to build their audience profiles, all panelists agreed that the marketers who are best utilizing the data they have, are those putting the most thought into how they measure it.