Getting to Addressability via the Transitive Property

By: Carla Holtze

We talk a lot about the fast pace of technology and the proliferation of data on these pages. And often, we toss around terms like transformative and disruptive to characterize all things digital. However, when it comes to real change, the pace can be downright glacial. For instance, we’ve been talking about brand safety for over a decade, yet it took an advertiser boycott of YouTube for that issue to be addressed in a more meaningful way.

And even general business trends sometimes take years to ripen. For example, data onboarding processes are not exactly new and took years to coalesce.  Data onboarding began nearly ten years ago with DataLogix. (Even longer ago if you include the ill-fated product that DoubleClick was attempting to build with Abacus). And while first-party data is currently all the rage, it’s worth looking back and noting that it took nearly a decade for first-party data to rise to the level of prominence that it currently enjoys.

More revolutionary ideas are fewer and more far in between. Apple’s IDFA was a revolutionary idea in part because it really only required one change agent (albeit one with a tremendous amount of marketplace power) that could impose its will on the rest of the marketplace. But for every IDFA, there are many other trends that take years to reach fruition.

One of the reasons for this is interconnected nature of the advertising technology supply chain. For every ad served, there are sometimes dozens of different entities that somehow impact that ad. From the publisher to the Supply Side Platform, to the exchange, data companies, verification and other participants from the demand side. Just to make this donkey parade we call digital advertising work requires tight coordination to say the least. So, when someone comes up with a truly revolutionary idea, it needs buy-in from well, nearly all of those donkeys. That’s a good deal of work, and a great deal of risk around every seemingly minor process change.

Making matters even more challenging, each of those touch points often needs to be in agreement that the change is in fact a necessary one. Fact is that where one sits within the adtech supply chain significantly impacts how you perceive the marketplace and your role within it. There’s a fairly complicated chess match at play here and an inconsistent level of trust. One person’s requested change “for the good of the industry” may be seen as another person’s change “for the good of their company.”

If third-party cookies were suddenly blocked by all browsers, we believe you’d see a groundswell of support for an alternative technology and a relatively quick transition. But absent some dramatic event that radically changes the game, real change is going to happen gradually.  But now, we’re seeing a few trends which are beginning to move the transition from a backwards crawl to a gallop. Perhaps we’re not at light speed yet, but these trends are significant enough to garner some momentum.

People-Based Marketing

The first is the trend towards people-based marketing. To understand this trend, one needs to accept the notion that people-based marketing will not be solely the domain of the walled gardens and in suit, will potentially be more cost effective for the marketer. As many transition from IBA based targeting to onboarding based targeting to people-based marketing, having an identifier that can be linked to a person in a “privacy safe” way (however that term is defined) is going to be critical.

EU Consent Requirements

The second trend is less of a business trend and more of a privacy and compliance issue. As you probably know by now, EU policymakers have created a rule beginning in mid-2018 that requires user consent that goes well above the current state of cookie consent boxes. We believe this marketplace change is the primary reason why Evidon jettisoned its Ghostery business – the market opportunity is significantly higher for building a consent mechanism than was presented by Ghostery. Similar to people-based marketing issued described above, one cannot build a consent mechanism without also building an addressability tool. It will be interesting to see if Evidon moves to create their own tool, partners with DigiTrust or some other vendor.

Device Graph 2.0

The third trend involves the maturation of the probabilistic device graph. Most device graph vendors currently rely upon a mix of probabilistic matches which are calibrated by a smaller “truth set” of deterministic matches. Depending upon the vendor, the ratio between probabilistic and deterministic may be as high as 9 to 1. That ratio was probably OK for device graph 1.0 – where scale was the name of the game. As we move to Device graph 2.0, accuracy becomes even more important. As a result, those same vendors are (or should be) frantically attempting to lower their ratio by raising their overall number of deterministic matches.

The Next Step for Advertising Technology

As implied above, part of the challenges inherent in effectuating major changes is a certain level of anxiety around “the new.” That’s just human nature. It’s almost always easier to borrow from concepts that people already have some comfort. And one of the specific challenges impeding the widespread adoption of the addressability graph has been the requirement that everyone flip the switch at the same time. Inserting a brand-new ID into the bid stream would require that all (or at least most) entities be equipped to use that ID on day one. If you have to get all the donkeys moving in concert towards “the new”, it’s going to take a while to get there. That was one of the challenges that held back the original DigiTrust so many years ago.

Accordingly, we believe that there will be a transitional phase where addressability IDs will simply be linked to the very pseudonymous identifiers that are currently used today. For example, linking a mobile advertising ID like IDFA to your own pseudonymous ID creates a common identifier that can be used to synch across devices and across multiple entities. Then all you need is one company (or multiple companies) to provide distribution of that ID to build out the footprint via the transitive property. And then rely on the existing onboarding vendors to make matches as they’ve done for years. We see this as the logical next step towards the buildout of the addressability graph for all of the above use cases. Whether this will be a six-month process or a multi-year process remains to be seen. But the business drivers are significant enough that we’re betting it will be closer to the former.

Carla Holtze is the CEO and co-founder of Parrable, a privacy-friendly identification platform. Previous to Parrable, Carla worked at the BBC, The Economist Intelligence Unit and at Lehman Brothers in both New York and Hong Kong. Carla earned her MBA at Columbia Business School, MS in Journalism at Columbia University and BS at Northwestern University. Carla serves on the committee for the San Francisco Symphony Symphonix, is an advisor to 500 Startups Data Track and is involved as mentor for emerging entrepreneurs and technology companies through Startup Mexico (SUM).

Follow Parrable (@Parrable) on Twitter.