Marketers Are About to Break the Big Data Bottleneck, But Then What?

Ray Headshot

By Ray Duong, GM, Optimizer at Lotame

Today, there’s no shortage of Big Data available to marketers. But for all the automation in collecting Big Data, marketers are still building and optimizing audiences manually through trial and error. The result is a costly Big Data bottleneck that undermines the value of a data-driven media strategy.

Breaking that bottleneck will depend largely on the continuing partnership between CMOs and CIOs, as well as important advances in machine learning tools. But while those tools are capable of automating and improving audience assembly and matching, the real breakthrough will come when empowered marketers deploy those insights throughout their organizations.

Adaptive optimization

Historically, marketing optimization has been a task with a beginning, middle, and end. A marketer uses research to come up with a hypothesis about their audience, they use data to target that audience, and then that targeting is optimized with the help of analytics and other digital tools. Unfortunately, every marketer knows that even when you optimize in real-time, there’s still a lag because consumer behavior is a dynamic, ever-evolving mosaic of data points. Adaptive optimization addresses this challenge through an ongoing Test, Learn, Evolve process.

Let’s borrow an example from recent memory to make this concept more concrete. A popular quick-serve restaurant chain suffers a massive health scare. For the brand in question, there’s an obvious product and PR challenge. But what about the other quick-serve brands in the space? Arguably, some consumers who were loyal to the first brand are up for grabs, but by the same logic, many more consumers might turn away from the entire category. Of the group of customers who walk away from the category, some may never return, while others may come back in a month or two when the uproar has died down.

Marketers have always been able to address this problem through data-driven insights, but that data has always come in hindsight. When CMOs and CIOs work together to achieve adaptive optimization through machine learning, that data surfaces in real time, and more importantly, so do the solutions. Going back to the quick-serve example, the other brands in the space can use adaptive optimization to determine which members of their core audience remain, which consumers are up for grabs during this disruptive period (as well has how to reach them), and which consumers are no longer viable prospects. Put simply, the Test, Learn, Evolve process is compressed to the point that the marketer is able to adapt as audiences change, not after they’ve already changed.

Adopt an audience-centric mindset

At the moment, marketing is all about drilling down on granular data to learn everything we can about specific audience categories like Soccer Moms. The trouble with this approach is that Soccer Moms don’t really exist. They’re marketing constructs, which means we’re using specific real-world data to illuminate marketing fictions. One of the key benefits of adaptive optimization is that it allows the CMO to take on a true audience-centric mindset where the focus is a real audience that is the product of data, rather than a marketing fiction.

Again, it helps to work with an example. Today, a brand that makes running shoes comes to their audience with an assumption—the people they think they want to reach are runners. Sounds like a no-brainer, right? Well, it isn’t. Putting aside the fact that runners come in a multitude of flavors, even the initial assumption is flawed. On the one hand, runners do buy running shoes, but people with lower back pain also buy running shoes because that particular style of footwear offers the kind of arch support they need.

The difference between these two approaches is night and day. By going after runners and using data to optimize around that audience, the marketer is essentially playing a trick on themselves; they think they have an audience defined by data when in reality they have an assumption supported by data. In the alternate scenario, where the CMO and CIO work from a data-first approach, they discover their true audience without being limited and led astray by assumptions. Obviously, only the second scenario offers the CMO what they really want: a way to grow business.

Data portability

In recent years, there’s been a push for brands to see data as an asset. At the same time, marketers have also been hearing about the need for data portability, which allows them to move between vendors and take their data with them to any media channel, inventory type, or content. But if marketers are truly going to think about data as proprietary (and they should), they need to consider what portability means in the context of adaptive optimization.

Currently, marketers ship out their data to DSPs and platforms like Facebook, but they aren’t getting real value in return. Or more precisely, they aren’t getting anything back that the CIO would consider valuable. Typically, what marketers get is a high-level report, and while it may contain insights, what’s lacking is granular feedback on the data the brand supplied. In effect, the CIO captures first-party data and the CMO gives it away. That’s a bad deal, and it has to stop. When the CMO and CIO are aligned under the banner of adaptive optimization, the CMO will understand that first-party data isn’t just something you spend within the context of a media buy, it’s an asset you enhance whenever you let it out the door.