Big, Big Data From Little Devices
Back in the early days of mobile advertising in the U.S., the carriers were the be-all and end-all of the industry. Before Apple and the app revolution effectively took AT&T, Sprint and Verizon out of much of the mobile media loop on smartphones, they were the keeper of the content keys. Getting on their coveted “deck” was every media company’s goal. And the nascent field of mobile advertising salivated over the prospect of one day being able to access all of that granular usage data the carriers supposedly sat atop. "Oh, the targeting we could do" was the reigning wisdom.
And so I will never forget a keynote speech that Jim Ryan, then head of then-Cingular, gave to a Mobile Marketing Association forum. He chided this conventional wisdom by assuring the marketers that, yes, Cingular and the other network operators had piles, mountains of data on users. But, to paraphrase Ryan, “it is in no shape that you would recognize.” Carriers weren’t thinking about phones as ad platforms much at all, except when they were afraid of alienating customers. They weren’t assembling this data in any shape that could be used by marketers, perhaps even their own, was Ryan’s point.
And while the app-based media economy doesn’t much need carrier cooperation the way it once did, it is still true all these years later that the operators have a level of understanding about mobile usage that no one can touch. I mean, they know who and where you are calling, for land’s sake. The mobile networks finally are starting to wake up to the big, big data they do have, but it is not likely they will be passing it on to third-party marketers before they themselves make the most of it. “There isn’t a carrier out there today who isn’t thinking about how to monetize their customer data assets,” says Lara Albert, VP of global marketing at Globys, a data solutions company that has been working with telecom as a software provider for over 15 years.
But getting into mobile advertising has always been merely on the periphery of carrier concerns. Their mother lode is the direct relationship with the customer over their basic wireless bill. Customer churn and the challenges of ARPU (average revenue per user) keep telecom marketers up at night. Yet, the tools they deploy in addressing these concerns often are staggeringly primitive. “You wouldn’t believe how many carriers still use profile data that change slowly,” she says. She and Globys are working with the industry to leverage transaction data in more intelligent ways that allow telecom marketers to target offers in real time. “Transactions equal behaviors,” she says. “You can look at how people behave in relation to one another. How do they vary according to context? And, predictively, you can know how best to act based on how they are going to behave.”
By harnessing the big, big data that cellular transactions create, the telecom marketer should be able to respond to user activity with the right offers at the right time. It is not always upselling but servicing and maintaining good customer relationships. One good use of data is to ensure that the customer is matched with the right plan, whether or not it is more profitable immediately for the provider. Keeping the customer happy and feeling they are getting long-term value could suppress the bane of telecom’s existence, churn. Keeping close eyes on usage behaviors and mapping them properly with the right offers to upgrade or economize on a plan keeps them with you.
Globys is working with some prepaid carriers to understand how and when people top off their wireless accounts with refreshed funds. They want to figure out how to encourage people to top off their accounts sooner and at higher amounts. “We are trying to understand it through contexts and offers,” she says. They may offer bonus cap off offers or loyalty rewards after specific behaviors. Albert says these are the kinds of marketing goals that require real-time behavioral data informing real-time marketing offers that are triggered by events. “They need to see what the customer is doing today - not who they think they are but what they are doing.”
It is the difference between segmenting based on profile attributes and using behavioral clustering. For instance, prepaid customers and their recharging patterns can be broken down into different styles. Some customers have spikes and valleys of recharging, neglecting their accounts for a while and then recharging high amounts on rare occasions. Others are following more regularly and in a steady flow. “You have to classify a customer by how they are behaving, not by how you think they behave based on some other profile attribute,” she says. Putting a customer into a profile bucket like a “low spender” is not telling the whole story. What if that low spender spends it all with you in a single day? “Static profile data may tell you a subscriber looks like this – a soccer mom, 40, with these interests,” she says. “But a richer longitudinal behavioral view may show she actually behaves much differently from others.”
The promise of using behavioral data in real time to make and customize offers and timing is that we use automation to get closer to one-to-one marketing. We may understand behaviors in clusters, but we market to people on a highly individualized basis. And then add to that the parameter of context. “You and I may have the same offer based on us behaving in different ways. But we may receive it in different kinds of contexts based on what we are doing that day or where we are,” she says.