A version of the following article ran in AdExchanger under the title: Which Telco Will Be The First To Challenge Facebook And Google? Below you will find an unabridged version.
The Looming Power of Telco
The Telcos are uniquely equipped for the battle for the people-based crown. They have the best mobile data, which is crucial for building the Identity Graphs that power people-based marketing. Telcos also have offline, validated, and billable people data. As a platform, you can have a mobile ID or an email hash and tie it to a cookie, or a device, or anything else, but rarely is anyone able to tie it to a real person that’s been “offline-verified”. Because of the billing relationship, Telcos have access to your phone number, credit card, address, and the number of rooms in your house.
Telcos, as I’ve argued before, stand at an advantage when it comes to the potential of dominating the people-based marketing era. Logins and log files are at the center of Identity Graphs, and telcos have access to logins, billing data and log files from their proprietary media at scale. All that Telcos need to do to create industry-leading cross channel identity graphs is fill in the gaps by adding algorithms and business rules to these vast sets of data like Telenor did when it bought TapAd. Rarely do pundits from our industry agree on anything, but those of us in the know agree that deterministic data is too sparse on its own.
Why Telcos Need Probabilistic
As an example, Verizon may know everything about my iPhone, but they only know a small amount about my iPad because I never use it to log-in to Verizon. However, the iPad is active 300 days a year alongside my iPhone. Additionally, I may have logged into Gmail on a shared computer and opened an email Verizon sent. In this case, Verizon has a deterministic match between my deterministic ID and that shared device. However, using deterministic data to target that shared device stops being accurate at the moment I log off. Probabilistic techniques used in conjunction with deterministic data would create a more complete picture that includes an understanding of primary (versus secondary) email, shared device, etc, which can help certain advertisers either restrict or expand of who gets targeted. This picture is only available to those that wield data at the execution layer.
In a world where everything is mobile and customers have been conditioned to engage with their phones, telcos could emerge victorious. However, the telcos still need varying amounts of expertise when it comes to technology, people-based marketing, and Identity.
That’s why AT&T’s hire of Brian Lesser from GroupM was so important. He can help AT&T, a neophyte in people-based marketing, acquire the technology they need to build out their capabilities. This puts them at the forefront of telcos making the plunge into people-based marketing alongside competitors Telenor, Verizon/Oath, and Singtel. Brian Lesser walks into a situation that is unencumbered by the data privacy restrictions that plague Verizon (thanks a lot, Turn zombie fiasco), and buoyed by the reputation of one of America’s oldest companies.
Telcos Will Seek Out Probabilistic Capabilities
As I’ve argued before, regions that went mobile first like Asia (Singtel/Yahoo Japan) and Europe (Telenor) are actually ahead of the pack when it comes to figuring out how to leverage Identity to make gains.
And what have foreign leaders learned? You need deterministic data at scale and probabilistic capabilities to tie enough devices together to enable true journey-based marketing, measurement, and attribution. This will create a new term for an industry already drunk on them: probabilistic people-based. The Identity Graph, at its core, is about mapping devices to intent data to login data and to people, which is no small feat in a world of where everyone uses multiple devices, browsers, and email accounts. Telenor pursued their Tapad acquisition because brands are insisting on personalized customer experiences that span browsers and devices. To enable “journey people-based marketing” it’s necessary to combine vast deterministic reach with probabilistic capability in order to solve a crucial issue for telcos: resolution across devices, channels, and platforms.
Think about your time spent in digital: much of it is in the sandboxes of app, video players, and email inboxes (sandboxes mean the cookie is only available in that application, unavailable to third parties). The problem is ad-tech is built on commercial cookies (third-party cookies available across websites). This is why wielding Identity is important: it’s the connective tissue that binds together the customer journey. There is fragmented data that needs to be merged together, and deterministic as the end-all be-all is misleading. The only reason we ended up thinking deterministic was the end-goal was because cowboys were able to claim probabilistic scale, and without having an universal model for measuring the quality of these probabilistic matches, the whole probabilistic graph concept was eroded. This is about to change, and Telenor (and others) have found out how to beat the scale of deterministic with probabilistic and still keep the same quality.
The Horse Race
I predict that AT&T, Singtel and Verizon will be the first behemoths in America to buck the trend of ignoring the power of probabilistic to tie together deterministic because they are in the media business and care about performance. They already have enormous deterministic reach. With probabilistic capabilities, they can double or triple their ability to resolve audiences as people cycle through their devices and validate the quality of deterministic data that’s important for both targeting and building the probabilistic models. They own the entire customer journey: from device to proprietary media to intent. The fact that Verizon, Singtel and AT&T have a closed loop means they can wield their proprietary data like Facebook and Google, and measure bottom-line success instead of leading indicators (accuracy; precision and recall). The very nature of processing data, building aggregates, and providing data-as-a-service is a double-edged sword. The data is easier to consume but like pasteurization, which sterilizes milk, it sterilizes data at the cost of the meta-data that is critical to probabilistics. This is why Telcos must own their own identity data and the associated meta-data. By doing so, they can examine the performance of the data and optimize for results. As such, I believe the move to acquire and activate probabilistic capabilities is imminent, and will signal an actual understanding of the challenges that people-based marketing solves.
As it stands right now, Verizon is closer to being fully operational for battle against the duopoly. While the telcos still own tons of data that connects to a real person, one weakness is the frequency of logins from primary email addresses. Most of the email addresses that telcos provide are not primary. I know that my Comcast address lives somewhere in the ether, rarely touched, never used for registering for an app or to sign-up for a newsletter. This is where Verizon sits apart. Verizon, by nature of their acquisitions of AOL and Yahoo, owns real, active email addresses checked dozens of times per day. Add to that their stack of Convertro (AOL), BrightRoll (Yahoo), ad.com (AOL), Adap.tv (AOL) and a leader in Tim Armstrong, a man who outmaneuvered Marissa Mayer with a business plan that was media-heavy and focused on content, not products, and they have an advantage understanding what really drives performance. Driving performance is the point of building identity graphs. Who will be the first to realize that the best Identity Graph is the one informed by an understanding of performance, not on metrics that aren’t relevant? Verizon and Singtel, with Amobee and Turn in tow, have a huge head start.
In a previous article, I had hypothesized that the Voltron that would come to dominate our industry had a high likelihood of emerging from the clouds. I now believe it will be a battle between the Telcos, the ascendant consultancies of IBM and Accenture, and Amazon which I will explore in a later piece.