WTF is a device graph?
Determining mobile advertising’s value in the marketing mix has always been tough. Measurement protocols are inconsistent across devices, and the main metric for desktop, the cookie, doesn’t work on mobile.
So, while people are increasingly accessing media across a range of devices, figuring out how much credit to give a mobile ad for a purchase has been a bit of a black hole.
That’s where the device graph comes in. While the idea (and tech) has been around for a few years, adoption is gaining steam and the tech becoming more mature. Facebook’s people-based (rather than device-based) marketing pitch, which relies on logged-in user IDs, has helped elevate the issue with marketers, according to ad execs.
Oh no, another ad tech term. What is it?
A device graph, also known as “identity management,” is a map that links an individual to all the devices they use, which could be a person’s computer at work, laptop at home, tablet and smartphone. Instead of counting each device as the behavior of a different person, a device graph counts them as one person, so there’s no duplication. Advertisers can then see things like what time of day a person was exposed to an ad and on which device, which helps show what role a mobile ad had in a purchase.
So it’s a turbo-charged method of attribution?
Yes and no. It’s dealing with the same issue, but it’s a more accurate method. There are two types of device graphs, deterministic and probabilistic. And marketers need to have both in the mix. Deterministic device graphs use logged-in data, such as when a person is asked to input their email address. Facebook uses this, as does Google, along with other companies like ISPs. Probabilistic graphs use location data to try and match the device to the person.
“Some vendors claim they get around 80 percent accuracy on this. It used to be 40 percent,” said ZenithOptimedia head of data and ad tech Samir Shah.
What other benefits are there?
Ad retargeting has gained a bad rep over the years for being too blunt a tactic. Being followed around the web by an ad for a product you’ve already bought is commonplace. But using a device graph lets you tailor the creative and the device the ad appears on, so it can appear at a more appropriate time and hence more relevant.
“If I’m looking at sports content on a desktop in the day, I shouldn’t then be targeted with shopping and fashion ads when on my mobile. That’s what device graphs do. We also want to ensure the mobile ad experience isn’t disruptive for people, especially with ad blocking on the rise. That’s why this is so important because we can connect their journey in a seamless way,” said ZenithOptimedia’s Shah.
Sounds useful for marketers
It is. But they’re not aware enough of its implications yet, and they need to be. “It’s critical to account for the gap in measurement there’s always been in mobile. It’s about the macro campaign experience across all devices. It’s become the device the majority of consumers are on. Marketing must catch up with consumer behaviors,” said Adform’s chief strategy officer, Anthony Rhind.
It’ll also give marketers more places to advertise outside of Facebook, said Blis’s vp of monetization, Dan Wilson.
OK. So what’s in it for publishers?
If advertisers can see proof of when mobile ads are contributing to a purchase on, for example, a computer, they’ll want more of it, so demand density will increase, which in theory will benefit publishers. “The CPM on mobile isn’t low because of the ad format; it’s because of the tracking. If buyers can link the value of mobile to the overall path to purchase, the price should go up,” said Rhind.
All sounds good. So what’s the catch?
For now, there’s no real way to account for device sharing. Many families share laptops and tablets at home, and smart TVs also are always shared. The key to link them all is the mobile phone, but increasingly young children have their own smartphone, or use their parents phones and tablets. That kind of throws all this out of the window, but for now, it’s the best answer the industry has come up with.
There are also some execution challenges and big differences between probabilistic and deterministic models. “Execution-wise, some advertisers or publishers may recoil from putting their data into a device graph to help enrich everyone’s models, but this is like self-driving cars — if one driver has an accident, that driver generally learns from it, and if a driverless car has an accident, every car learns from it. It’s the same for a device graph,” said Blis’s Wilson.
Click here to see the original article.
Tags: Dan WIlson, Device Graph, Digiday UK