It was a good week to catch up with Mike Baker, CEO of data management firm DataXu. After founding the firm a few years ago, Baker has built it into what Forrester said this week was among the top DSPs in a tight cluster of larger similarly rated firms, MediaMath and Turn.
Forrester’s analyst Joanna O’Connell called the company “an up-and-comer to watch.” Baker took a curious path to being a Web “up and comer.” I knew him for years in the mobile world as the leader of an early mobile ad tech company and network EnPocket. He sold the company to Nokia in 2007 and soon left what was then the 800-pound gorilla of mobile, before Apple entered the game.
I always wondered how he made that pivot from mobile back into data. “The iPhone was just hitting the market and you had a sense just being in mobile since 2001 that a chapter was concluding in the evolution of the consumer’s use of mobile,” he says. “It was a good time to take stock.” The usual non-compete agreement also helped get him thinking outside of mobile circles.
Baker says that much of the vision for building a larger data management company came from one of his last endeavors at Nokia, to project what the media world would look like globally in 2020. In that project 90% of all media was seen coursing through digital system, producing massive amounts of data and bringing the need to make an unimaginable number of real-time decisions about what and where to serve -- to whom.
“This was Nokia, so you had to think globally, about Sub-Saharan Africa and everywhere on earth," Baker says. " In that world when everything is digital there will be trillions of dollars in marketing and advertising. The volumes of 8 or 9 billion people on digital devices and all of that data will be one of the defining characteristics of 2020.”
After leaving Nokia, Baker started talking with MIT teams working on decision science and how to handle billions of possibilities in a system. They were working on solving unprecedented problems by creating a decisioning system out of multiple algorithms and databases. In an early test of their system in modeling manned space flights to Mars, the MIT team bested systems from Lockheed and Boeing, Baker says. The system offered up thousands of different architectures for a mission to Mars around a large range of different priorities.
DataXu essentially was taking the core abstract modeling science out of theory and putting it into marketing. “We were trying to get some of this math out of the academy and apply it to business problems. We said, let’s write software and solve some marketing problem.”
Launched in 2009, DataXu started generating revenue early, and this year tripled its sales over last year, Baker says. He positions his DX3 data management platform as providing 20/20 vision: “a decision system sitting above all the chaos and point solutions that have been guiding investment and tells the systems what to do.”
The DX3 system has connects to over 22,000 data segments to forecast media planning. “Now a buyer can see the cost of media,” Baker says. And in a digital world currently dominated by Google, he argues the need for a management platform that is independent of the sell side so that buyers and brands can create, test and manage their own models. “He who owns the algorithm takes the lion’s share of the economics,” he says. “But what if big brand marketers, the finders of these companies, what if they owned their own algorithms? The buyer strikes back – that’s our premise.
In Forrester’s comparative analysis and critique of the top DSPs, O’Connell highlights DataXu’s strength in attribution, a feature Baker understandably emphasizes as the path forward. “No one has cracked the code,” he admits, but a higher level view of data across inputs and outputs allows for buyers to get beyond post-facto analysis informing the next campaign.
“If you are doing split A/B testing to see which products give the best return then it becomes a buying model,” he says. “You can create a custom set of rules to see how you best deal with someone who saw a video and then a banner and then clicks on a text link.”
He is betting on the need ultimately for that master console that diminishes the distinctions among the many specific solutions the digital ad industry has spent the last decade generating. “When this industry matures, it won’t be rich media or behavioral targeting or DSPs and DMPs. These are all the little threads going on.”
Baker also cites an interest in moving some of the efficiencies buyers see in ad exchanges and DSPs to more premium inventory. An important new project for DataXu will be letting buyers apply decisioning against guaranteed media buys so a marketer can optimize creative on the fly against different audience segment on the page.
But ultimately, even four years out of Nokia and his EnPocket days, Baker suspects that mobile is the next big hill, but it is also the place where all of this data at last closes the loop. DataXu already has mobile products, but he admits that buying audiences DSP-style across such a fragmented landscape is heavily constrained by where a cookie can travel. "The good news is, we’re seeing great results just from doing optimization of mobile advertising,” he notes. The company has been running tablet ads that optimize off of conversions on straight contextual buys that are delivering what he describes as “eyebrow-raising” CPA efficiencies.
For mobile the data point everyone is waiting for is the end sale, of course. When all of this data that DataXu and the others in the data management space have been consolidating finally can be tied to a sale, then all of the spending can be optimized against the ultimate data point.
Mobile will provide the offline signal that the online world needs, and so the stakes on these little handheld devices are higher than just mobile ad budgets.”2012 will begin the mobile payment battle,” says Baker. “The big catalyst for mobile is getting the offline purchase signal via the device. That represents a capability that surpasses online.” And perhaps brings the former EnPocket/Nokia executive full circle.