ID Consortium Consolidation

There are currently three ID consortia in ad tech: DigiTrust, the Advertising ID Consortium, and The Trade Desk's Unified Open ID. It was recently announced that DigiTrust and the other two would allow for interoperability. We discuss the nuances and impacts in this week's Lift Letter.

LiftLetters-IDConsortium.jpg

DigiTrust originated in the IAB Tech Lab, was spun out in 2014, and then re-absorbed in April of 2018. DigiTrust uses the digutru.st domain and includes BidSwitch, Kargo, Sovern, and Quantcast as members. The Advertising ID Consortium was a combination of various ad tech vendors. It was based on the LiveRamp user ID, hosted on the AppNexus cookie domain (adnxs), and supported by a large number of industry players, including Index Exchange most vocally. Finally, The Trade Desk created the Unified Open ID, hosted on its own domain, adsrvr.org

The ID consortia solves a real problem. If you assume there are 20 SSPs and 20 DSPs that matter, and that each DSP and SSP has its own unique user ID, then it takes 400 cookie syncs for each SSP to know each DSP's IDs for every user. In a little more detail, because the SSP only sees cookies for its own domain on requests made to that domain, and because the DSP never actually accesses the cookie in an ad request (it only sees the bid request from the SSP), the SSP needs to know all of each DSP's IDs. The SSP gets this information through cookie syncs where a "pixel" request will forward a specifically formatted call from the DSP to the SSP, so the DSP tells the SSP in the URL portion of the pixel call that user's ID, which the SSP then maps to that cookie ID in its systems. SSPs and exchanges like TripleLift store mappings for every user in high-performance mapping tables, where the right user ID for each DSP can be included in every bid request.

There are several types of costs associated with the requests above. First, in TripleLift terms, the more mappings TripleLift has to store, the bigger its mapping table has to be. Exchanges need to do this lookup on every single ad request, and it needs to be incredibly fast. This means the technology here is sophisticated and expensive - and the more data that's stored, the more it costs. Second, in monetization terms, each time there is a new user, only a fraction of the 400-or-so requisite cookie syncs will have completed on any given page load into that user's history. When you factor in other companies like data providers, attribution vendors, etc. the number of required syncs can be in the thousands. This means that some DSPs will not be able to bid with their complete set of demand - where the appropriate syncs have not occurred, thus the monetization for that impression won't be what it would be in an equivalent world where all syncs have already happened. Because cookies statistically degrade over time due to a variety of reasons, it is not the case that this simply sorts itself out and goes away - inadequate cookie syncing may be a significant factor for an exchange under-performing overall (being the first cookie sync for a user may have an lasting impact). Finally, cookie syncs may simply interfere with the user's experience with a site. While a single sync is generally inconsequential, some sites issue dozens or even hundreds, which may cause a degradation in overall performance and lead to less time on site and lower overall ad monetization.

Solutions to these problems are in the interest of nearly all players in the ecosystem. The general premise of each of the cookie consortia is that a single cookie domain, with a canonical ID, sits at some point in the ecosystem. One could imagine a system where a third party stored its own "trusted and independent" ID. TripleLift, and every supply partner, would sync with that vendor. Similarly, every DSP would sync with that partner. This would require 40 as opposed to 400 syncs. An SSP could then send the independent partner's ID, as opposed to each DSP's own mapped ID, to the DSPs. This would mean fewer lookups (or less total lookup data), fewer syncs, and a better user experience. Not every consortium works precisely this way, but it's the general gist.

Prior to the IAB acquiring DigiTrust, each of the three consortia were for-profit and owned by participants in the ecosystem. Further, there were obvious conflicts of interest. AppNexus, for example, is the owner of the adnxs domain, an SSP, and also a DSP. When considering SPO, it would be naturally at an advantage to bias towards its own SSP, given it would necessarily have the highest match rate. Similarly, The Trade Desk - while not as obviously conflicted - would create advantages in the ecosystem against other DSPs. The decision to incorporate the DigiTrust domain into the cookie consortia specifically addresses the concerns around the potential for conflicts.

Ultimately, the final goal of the participants is beyond simply reducing the costs mentioned above. Once a vast number of industry participants have settled on a single ID for an individual, that moves away from relying on cookies, and creates the potential for relying on more persistent identifiers (e.g. email) when creating - or recreating - the cookie ID. Ultimately, the design is that through this persistent, cross-device ID, ad tech participants can begin to affect people-based marketing of the sort that only the walled gardens could historically provide. Of course, whether this particular part of the strategy survives the current trend towards ever-increasing privacy remains to be seen.