This week , I interview Brendan Riordan Butterworth. We exchanged emails for this interview and it has been slightly condensed and edited.
Marc Goldberg: We first met when you were at the IAB, tell us a bit about yourself and what you are doing now.
BRB: When I joined IAB, I had to look Randall Rothenberg in the eyes and disclose my participation in the creation of Adblock Plus, way back in 2005, when I was roaming Australia. I spent the latter half of 2005 working to convince the web analytics company I had resumed working for that we needed to develop data collecting technologies that were resistant to the capabilities of blocking tech.
MG: So you helped create adblockers? If you had looked me in the eyes, I would have punched ya! What are you doing these days?
BRB: Yes, fortunately I had a few years with Microsoft under my belt before that conversation. These days I’m still trying to stay ahead of the curve – consulting for a few companies that are building for, or at least trying to influence, what happens next for delivering advertising efficiently. And while there’s lots of interest in maintaining ID graphs, some other techniques are of interest as well.
MG: So an ID graph, is that what the guy that used to create my fake id 100 years ago used?
BRB: With a drinking age of 18, and a cultural expectation of kids shopping at the corner store for their parents, fake IDs were not a thing for me growing up in Quebec, Canada – so I’m not sure what your guy was using.
Identity graphs these days are used to bridge recognition of a user across contexts.
In the world we’ve advertised in since forever, this has been done by cookie matching. However, with third party cookies going away, identity graphs need to move to using alternative sources of information – like primary and secondary email addresses, shipping and billing addresses, device concurrency, payment information and emailed receipts, and and a variety of other deterministic and probabilistic models of associating a set of different IDs from different contexts with a single user.
If that all sounds complicated, it’s because it is – losing access to third party cookies introduces significant challenges to the way digital advertising has been done.
MG: So where do you see Identity in 2023 and what does everyone have to do now?
BRB: Don’t Panic.
In 2023, I expect to see a small set of individually trusted publishers operating with a high percentage of registered and authenticated traffic, and a medium set of trusted publisher conglomerates operating with a smaller percentage of registered and authenticated traffic. Current estimates are that this might end up representing 15-20% of total impressions, though that might be a touch low.
The remaining three quarters of impressions won’t have a unique user identity.
There’s work being done in Google’s Privacy Sandbox arena and the W3C to offer cohorts – browser-managed identities for groups of users with the same set of interests. This is only just “taking off” – the effort is still a fledgeling one.I’ve also worked with eyeo GmbH to propose a totally client side profiling system, and we’ve now shipped a product called Crumbs that collects all the privacy features users are looking for and provides them in exchange for running a local ad profiling system that determines membership in segments and shares these directly with advertising systems, without user IDs or cohort IDs.
MG: What the FLoC are you talking about?
BRB: Exactly – moving away from 1:1 advertising and towards cohorts, one way or another. We’ve recently seen the Google Ads team claim that simulations of the FLoC algorithms suggest that the cohort based methodology can result in “at least 95% of the conversions per dollar spent when compared to cookie-based advertising” (ref).
Personally, I have concerns – since cohorts are aggregations of users, they’re unwieldy for handling typical sequential advertising scenarios that we’re accustomed to doing at the user level, like retargeting, suppression, frequency capping, and attribution, among other things.
While cohorts represent significant change, the move to registration or authentication in order to get hashed email changes the conversation a publisher has with it’s users from “click OK to accept cookies” to “let’s talk about sharing your email address”. We’ll see some publishers unwilling to go there, and some users uninterested in doing so. I’m excited about how the variety of potential approaches will cover the different levels of comfort that individuals users have with different publishers.
Back to specific actions for 2023 – publishers will have to know their customers, and the value of their audience more directly, and the value that they bring to their audience.
Advertisers will need to understand how to use context, and to establish the systems to scale trust from the relatively small set of third party advertising and measurement systems to a more distributed model.
And everyone will need to keep an eye on what browsers and other user facing tools are bringing to the table.
On the emerging technologies front, I hope to see some of the threads of a provably trustable network start to come together. For example, certifying the attributes of creative payload, like what Cloudflare and Google have done with Firebolt and AMP for Ads or using the Ad-ID infrastructure to know that it’s really an ad from Coke. Also looking at what Brave has done with BAT and figuring out whether that’s a viable foundation for proving measurement in private environments. Even determining how to scale trust in content by including cryptographic certifications attesting to its origin (like “this publication is recognized to adhere to these journalistic standards”) or context (like “IBM Watson determines these categories are relevant to this text”), so that advertisers and other systems can make better decisions about what and when to serve.
BRB: Google, Facebook, and Apple are all decently positioned to ignore, for the most part, the impact on business from the change in cookie rules. Google will have to make good faith efforts on the Privacy Sandbox initiative due to the ongoing legal challenges to deprecating the cookie.
Large, trusted publishers with a significant portion of inventory being delivered to signed-in users, especially if those users are subscribers, are really well positioned to continue business as usual.
They’re also positioned to grow, by leveraging knowledge of their own content to optimize against a growth of demand for contextual buys. Look at Hearst aspirations with Mylo, for example.
Medium-sized or local publications that have sufficient audience trust and can coordinate to establish reasonably sized pools of authenticated traffic will do OK. They’ll probably not directly implement UID 2.0, but do this through a partnership with an identity system that leverages this.
MG: Who is screwed?
BRB : This is going to be a hard time for publications that depend on short interactions with visitors.
Any DMP that can’t make a pivot to also being an ID graph provider. And any new ID graph provider that can’t establish enough partnerships to be competitive.
Personally, I hope that last touch attribution and the intensive retargeting that’s used to game that model get screwed.
MG: Are marketers set up for the new normal? Expectations of digital targeting could be at risk , will they be reluctant to grow this digital pie?
BRB: The first step of digital targeting is being worked on extensively, and I expect a reasonably smooth transition – marketers will still be able to define segments and reach them. It’s everything that follows that’s at risk – the ability to retarget, measure conversions, deliver a series of ads and be sure that each user saw them in their intended sequence.
There’s some talk about marketers going back to media mix modeling methodologies from 50 years ago – that’s just not feasible.
Nielsen is doing some interesting work that I interpret to be extrapolating observed behavior of their panels to larger data sets, and I think we’ll be seeing the application of this type of methodology to handling the decreased data resolution for the large swaths of not-directly-addressable inventory.
In my opinion, 2020 has pushed society nearly a decade ahead in terms of digital literacy. My dad has started sending me memes! It only makes sense to continue delivering marketing digitally – even if addressability is down, attention is way up.
MG: You have twins four year olds, which one do you like better? Answer that and I drive to brooklyn and hit you in the back of the head. You have three young girls. 7 and the twins. How has quarantine been for you with them?
BRB: I have each of their names on a sticky note, and I reorder them daily. Technically, it’s for their seats at the counter for meal time, but whoever’s sitting in the middle does get more attention.
The first few months of lockdown were really hard on our family. We ultimately joined 3 other families to form a pod with 4 kids in each age group, which has been a total game changer. Consulting primarily for a European company also has meant great afternoons with the kids, and we’ve taken full advantage of our proximity to Prospect Park.
MG: Are they in school?
BRB: All remote – and, with the pod, they’re mostly learning independently and just checking in with the core competencies as needed.
MG: If you had the chance to dine inside , what restaurant would be first on your list?
BRB: I love the classic diners – shout outs to Cobble Hill Coffeeshop in my old neighborhood, and Purity Diner in Park Slope.
While I wait for the legit experiences, I host Friday night pizza on the regular, thin sourdough crust and precisely weighed fresh mozzarella cheese, and Saturday morning buttermilk pancakes or sourdough waffles. I’m also making sourdough bread 2-3 times a week, crumpets on the regular, cookies, zucchini bread, as well as fermenting buttermilk, hot sauce, sauerkraut, and anything else that seems like it might get tastier sitting in a slightly acidic brine with lactobacillus colonies for a while.
Given that the majority of my patrons are between 4 and 7 years old, I’m getting quite used to scathing critiques and stingy tips.
MG: We don’t have similar hairstyles. Did you get a haircut in 2020 or plan to get one ever?
BRB: None in 2020. I did briefly consider getting a Flowbee. These days I’m slowly shrinking my quarantine beard – no sense in traumatizing my kids with a suddenly shaven face.