With ATS Paris nearing ever closer, ExchangeWire hears from the upcoming event speakers, to give a flavour of what to expect on the day. Here, ExchangeWire speaks with Jakob Bak, CTO and co-founder, Adform, and Anthony Rhind, CSO, Adform (both pictured below), about that elusive challenge of bridging the cross-device gap, ahead of Rhind’s participation in a panel discussion covering just that in Paris on Wednesday.
The ATS Paris panel dedicated to cross-device is arguably the most important for the endurance of ad-funded editorial business models. Without significant changes to digital ROI measurement methodologies, publishers’ audience monetisation will continue to be damaged by people shifting content consumption to mobile OS devices.
With internet-enabled multi-device ownership now pervasive, marketers need reliable cross-device mapping to inform analytics and organise activation for both media and creative. Adform and others have invested heavily to develop this capability; but even deploying effective mobile OS device-level tracking is not yet an industry standard. Publishers are feeling the pain of this most acutely, but agencies and advertisers must address this before cross-device mapping can be deployed effectively.
For a tech stack like Adform, this has meant developing highly sophisticated device-level data capture first, which contributes to making development of cross-device solutions something requiring significant investment of brainpower.
The core cross-device mapping challenge, driven by the proliferation of increasingly mobile OS devices used by a single person, or shared within a family, actually echoes the challenge created by cookie deletion: to consistently recognise actions and data for a single (anonymised) person.
Unfortunately, establishing an accurate association for devices is complicated. In our experience, there are three major and distinct challenges: data access, mobile identification, and technology upgrade.
Challenge one: Data access
The first and most obvious challenge is establishing access to a universal data set. Even Facebook have neither the majority of the internet population logging in, nor every user accessing Facebook on every device they use. To reach a data volume that offers a near-complete view of the devices used by most of the population means going beyond a single dataset. For a company like Adform, the only solution is to aggregate as many deterministic datasets as possible in our DMP, then augment with several probabilistic cross-device data providers.
Advances on the probabilistic front certainly represent a strong positive as the industry moves cross-device solutions forward.
If we look at our initial experiences early last year, the reach and accuracy of the probabilistic graphs provided by Screen6, Tapad, and similar specialists have evolved significantly.
The reach is far above what we would estimate even Facebook or other global publishers to have. We initially feared this would not be achieved for several years, or indeed without contribution from deterministic sources such as the Telco’s. Instead, we’ve seen accelerated development, which is incredibly welcome as it means that several of the key use-cases for cross device graphs, such as targeting and audience extension, are already possible via Adform and other leading DSPs.
Unfortunately, probabilistic graphs will probably always struggle to establish the reach necessary for perfect frequency capping and to feed attribution analytics. Therefore, the aggregation of deterministic data continues to be a necessity.
At Adform, we carried out our first cross-device login data/DMP collaboration with a publisher about a year ago. Since then, we’ve found that it is an extremely slow and difficult decision process for publishers and other media-related data providers to commit to a next step. The interest is there, but it seems that many publishers are hesitant to be first movers.
We have experienced better traction on the advertiser side, where large e-coms and similar players with significant cross-device logins have proven much more willing to share this data, provided that they gain access to the full graph for usage in their Business Intelligence or on-site experience management.
While the availability of deterministic data directly from publishers and advertisers will increase, we hope (rather than predict) that this year will see the emergence of significant deterministic datasets via the leading data exchanges to supplement the current limited supply.
Challenge two: Mobile identification
After data access, the second big challenge is of a more complex and technical nature. Mobile OS device identification has been a key pain point since early smartphones. The ability to deal with non-cookie in-app environments, as well as cookie limitations in mobile-web, is a key problem when it comes to doing cross-device correctly. It is worthless to have login data on desktop, mobile web, and mobile app if this data is invalid by the time the user is seen!
Another aspect is the challenge to be able to activate cross-device data equally in all three environments: namely desktop, mobile web, and mobile app. At Adform, it has been natural for us to invest heavily into our mobile ad server and mobile DSP capabilities because the platform is used as a single stack solution by most of our clients. As mentioned earlier, much of the industry has failed to implement effective mobile tracking; without this, the device graphs lack both historical communication context and the tools to activate a connected communication experience.
Challenge three: Upgrading technology
Last, but not least, is the significant challenge of upgrading technology to support a new multi-ID world. For us, it has been a colossal two-year challenge to rebuild all core infrastructures so that targeting, frequency capping, and attribution run on a multi-ID backbone. This rebuild was certainly complicated by including almost all development teams and needing to address all of the existing code base. Building a standalone cross-device DMP or DSP would naturally have been much easier. The slow emergence of fully integrated cross-device support in similar stacks or standalone DSPs, DMPs, and ad-servers indicates that we have not been alone in finding this a significant and resource-intensive project.
The future is now
At Adform, we still have a lot of additional functionality ideas we intend to build; but our immediate focus is to court partnerships with deterministic data providers. At the same time, our priority is to drive uptake by agencies and advertisers. It is an exciting time as the basics are now in place to witness the initial benefits associated with cross-device mapping and activation.
Once we bridge the ‘digital media’ gap, there are a number of exciting opportunities, one of the most compelling of which is TV. Connected TV’s will soon be added to the device graph. Since they are IP-based, it will become possible to determine what TV ads a household is watching, then map to the household member’s device graphs based on a combination of device proximity, time-based, and content behaviour patterns. Including, of course, measuring TV ad contribution to conversions alongside brand awareness KPIs in real-time.
There is a lot of friction to reach that nirvana, we’ve already acknowledged that cross-device is an onion that gets more complex as we peel layers back. But, we are talking regularly with fascinating, forward-thinking, companies and eager to push what is possible through programmatic TV.
Rhind will be participating in a panel discussion on ‘Bridging the Cross-Device Gap’ at ATS Paris on 13 April. Find more information here.