Advertisers are understandably looking for scale with their cross-device campaigns – but when they ask for it at the expense of accuracy, that’s a problem.
“The marketer has erred on the side of reach, but reach and frequency are a tradeoff,” said Omar Tawakol, SVP and GM of Oracle Data Cloud, speaking at a panel during AdExchanger’s Industry Preview conference Wednesday in New York City.
In other words, pretty much any cross-device identity graph – other than, perhaps, Facebook’s – is a compromise between getting it right and getting it right at scale. And for cross-device recognition or targeting to be successful, you need both of those cylinders firing.
But advertisers often evaluate vendors based on reach because it’s an easier metric to grasp.
“Just give me a number, tell me how many devices you can give me when I give you a set of cookies – it’s really easy,” Tawakol said. “But when you talk about accuracy, people don’t even really understand how to prioritize [it] or what truth sets they’re going to use.”
While Facebook can use deterministic data to link multiple devices to individual users, other cross-device vendors in the space generally rely on probabilistic techniques to create their connections.
Of course, advertisers can also bring their own first-party data to the party – LendingTree, for example, recently created a login experience to collect registration data and get insight into the customer journey, said Nitin Bhutani, VP of marketing at LendingTree – but scale is an issue.
In a probabilistic setting, where matches are being made using statistical algorithms, expanding reach can mean sacrificing accuracy.
In general, though, marketers have “to get smarter [and] ask questions about accuracy,” or they’re going to “force the wrong behavior with vendors,” Tawakol said. “Think about what your vendor is going to do in the cross-device space if the next time you meet with them you say, ‘Well, this other vendor has slightly higher reach.”
If marketers demand reach, that’s what the vendors are going to provide them with.
“Well, I’m going to start a device ID graph company tomorrow and say everyone on the planet is connected – maximum reach,” Tawakol joked. “It’s ridiculous.”
That said, reach and accuracy don’t necessarily have to sit on opposite ends of the seesaw if vendors use a combination of deterministic and probabilistic methods, Tawakol said.
‘We all want to use deterministic, but there is nowhere near enough matches to do that,” he said. “We need to do more of a data science approach that takes what you know and scores it probabilistically.”
But what about Facebook and Google? They’ve got serious deterministic mojo, obviously, but the rest of the industry isn’t licked.
“Their scale and reach at this point is unbeatable and the amount of verified registration data they have is top notch, but they don’t have full coverage for cross-device yet,” said Jennifer Lum, co-founder and chief strategy officer at Adelphic. “[And] they have work to do in TV and on the mobile web.”
None of the debate around consumer identity matters, though, if brands don’t consider what they’re going to do once they’ve created the cross-device match.
“All of the companies we work with have varying levels of fantastic first-party data and access to third-party data and, increasingly, we’re helping brands work together to generate significant second-party data,” said Matt Asay, VP of mobile at Adobe. “But, stepping back from that, the most important thing for any brand is not, I’m sorry, which ad you show the person – it’s the overall experience that consumer has with that brand.”
