Rocket Fuel Panel Tackles Big Data Opportunity

tackling-big-dataFiguring out how to leverage the overwhelming amount of data they collect is a critical challenge for advertisers, and one that spans industries. To shed some light on big data issues, real-time ad platform Rocket Fuel invited Forrester Research analyst Mike Gualtieri, Visible Measures CEO Brian Shin, Morpheus CMO Shenan Reed, and Verizon marketing and business development director Tanja Omeze to discuss the challenges advertisers face in making sense of their data.

Visible Measures’ ultimate goal is to “identify causation,” explained Shin. “We track online video viewership and part of what we do is look at how people respond to different types of advertising,” he said. “First we try to recognize patterns, determine correlation…and if you can understand why something is happening, that’s really cool.”

In contrast to Shin, Gaultieri stressed the difference between correlation and causation. “Everything I talked about [in using big data] is not causation—it’s 100% correlation,” he said. “We’re not looking for evidence-based methods. If you have evidence-based methods for predicting something, you don’t need data science.”

Rocket Fuel CEO George John asked the panelists the following questions, “How do you see the world changing? Are your customers more attuned to the opportunities behind their data?”

Reed’s response: “The desire to get to the individual is not new. I just think we have more processing power to make it happen. Especially when we’re dealing with people already in a digitally connected space, you have more connectivity.”

Omeze said, “There’s also been a shift in people’s comfort level for sharing data. People are starting to understand that they can get value in exchange for information.”

At a later point in the discussion, in response to an attendee’s observation that many ads continue to be poorly targeted, Gaultieri noted, “All media is set up to be pushed. Advertising has to become media where consumers can choose what they want to interact with. Choice by definition equals content and not advertising.”

Another way to look at advertising is as a service, Reed added. “Advertising in its purest form is a service,” she maintained. “It gives you the right information at the right time to help you make a decision or create a moment of serendipity.”

Another attendee argued that targeted advertising reduces the opportunity for consumers to be surprised by a product or service. “Aren’t we actually narrowing our choices in terms of what’s offered to us? I don’t want to know everything in advance. I still want the illusion of free will,” she said.

Reed agreed. “That’s also the biggest concern for advertisers,” she noted. “If my clients put everything into targeted advertising, then you’ve self selected a group of people that have the highest probability to buy, but taking over the homepage of The New York Times during Fashion Week shouldn’t go away. Those are opportunities for discovery. If you lose that discovery you end up with siloed buckets of people.”

According to Gualtieri, a data scientist’s answer to the attendee’s question would be to predict a person’s propensity for serendipity. “You just need to find a balance. Give everyone a propensity score and serve them that way,” he suggested.

After the discussion, AdExchanger asked Rocket Fuel’s John to speculate on what the panel’s discussions would be like five years from now. Very little will have changed, according to John.

“Even in the 90s, we had data mining and data warehousing, telcos had churn models and we had cell phones,” he said. “The kind of value that gets added by analytics and data scientists is the same from a long time ago. These are massive problems companies are tying to solve: huge databases, integration issues, human interpretation. All of this is very challenging so, if we did something like this [panel discussion], it would sound the same.”


Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>