See the list of my annual predictions posts, dating back to 2016, here.
Of course, mobile advertisers knew from the earliest days of ATT that the policy would significantly burden advertising performance and foundationally upend the dynamics of the mobile ecosystem. ATT was a predictable eventuality: I discussed the inevitability of a policy like ATT in 2017 in The coming war between Apple and Facebook.
It’s important to consider ATT in the context of the broader tectonic shift related to privacy that is taking place across digital platforms. ATT did not spontaneously materialize in a vacuum. And while I see ATT as a competitive campaign by Apple to re-center the nexus of content discovery and distribution within the App Store away from ads platforms like Facebook, and not as a genuine consumer protection as Apple contends, I believe the undercurrent of skepticism and distrust around how digital platforms utilize consumer data will ultimately lead to a more productive and more mutually agreeable relationship between digital platforms and their users. Increased scrutiny of the collection and usage of consumer data — from platforms themselves, and from governments — will alter the landscape of not just digital advertising but of our digital lives in the years to come. The privacy landscape is irrevocably changing and advertisers must adapt to those changes. ATT is just one component of that change; privacy concerns will moderate the operational mechanics of mobile advertising from every direction.
It is against this backdrop of radical upheaval that I present predictions for 2022. This year, I am making fewer precise predictions, opting instead to offer what I believe are the broader trends that will dominate the mobile marketing ecosystem over the course of the year. Because of this, the title of this post is a bit misleading, but for the sake of consistency, I kept the wording the same as all past predictions posts.
I find the exercise of writing predictions each year (as I’ve done on this website since 2016) to be very helpful in sharpening my own thinking: even if the predictions are wholly incorrect — and often they are, to which past years’ predictions posts attest — the process of looking at mobile advertising from a higher altitude and attempting to extrapolate out a course for the next year is illuminating. At the very least, it forces a wider lens to be used in considering what is happening right now. That said, below I outline what I believe will be the four most meaningful trends in mobile advertising in 2022.
Proliferation of Content Fortresses
I coined the term Content Fortress last year to describe a platform that services content interactions for third parties within its own first-party environment, allowing it to use that data for the purposes of ad targeting. As I explain in the tweet thread below, the key distinction between a Content Fortress and a Walled Garden is that the former creates value by keeping the user within its product (servicing engagements), the latter creates value by forwarding the user out of its product (servicing ad clicks). The canonical example of a Content Fortress is FB Shops (and the new ecommerce products that have been rolled out at every large social platform, such as at TikTok and Twitter). The FB Shops product allows merchants to host their stores and service transactions directly within the Facebook Blue and Instagram apps: because users never leave a Facebook app, Facebook has full transparency into those transactions in a first-party setting and can use purchase data to target ad campaigns in compliance with ATT and, likely, upcoming privacy regulation.
Common question I field: “What’s the difference between a Walled Garden and what you call a Content Fortress?” Short answer: Walled Gardens get paid when users leave, Content Fortresses when users stay (1/X)
— Eric Seufert (@eric_seufert) November 11, 2021
The Content Fortress will be a dominant theme in 2022 as the largest consumer tech platforms attempt to augment the scope of their content reach and user engagement touchpoints and expand the surface area of their first-party purview, as I describe in Everything is an ad network. This can be achieved through two directionally-oriented strategies: ad networks buying content companies, or content companies buying ad networks.
PayPal’s rumored potential acquisition of Pinterest would have created a formidable Content Fortress, pairing PayPal’s considerable pool of purchase data with Pinterest’s ads platform: the unified data set could have been used to serve targeted ads to Pinterest users based on behaviors and interests gleaned from their PayPal purchases. These kinds of mega-deals are by definition rare, but M&A activity on a smaller scale supporting the Content Fortress thesis has been frenetic in the wake of ATT and will only accelerate in 2022. I think this trend will be particularly prominent for companies that traditionally lagged on mobile. These companies will see an opportunity to make up lost ground through M&A as a result of the disruption caused by ATT and forthcoming privacy regulations.
Read more: https://fastento.com/
Battle for measurement authority
With the digital advertising privacy environment fundamentally subverting the “hub-and-spoke” model of conversion measurement that I detail here and here, ad platforms are incentivized to aggrandize their measurement authority, either by offering “single source of truth” tools like Media Mix Models (such as Facebook’s Robyn tool) or by simply modeling more conversions than they did previously.
It’s easy to understand why Facebook is aggressively evangelizing Robyn, its tool for media mix modeling, if you understand what’s happening in the iOS mobile advertising ecosystem: conversions from O&O properties are being dramatically undercounted because they use SKAN (1/X)
— Eric Seufert (@eric_seufert) November 23, 2021
In 2022, I expect large ad platforms will begin to make earnest efforts to route all advertising performance measurement through their systems. This will be sold to advertisers as a means of simplifying measurement: who better than a large advertising platform to build these complex systems? Prior to ATT, mobile advertising measurement was outsourced to middleware attribution companies, except that “self-attribution networks” like Facebook, Google, Snap, etc. captured their own conversions and reported those to advertisers. I discuss why the notion of deterministic measurement was mostly specious prior to ATT for this reason in this podcast with Maor Sadra, and I spell out the distinction between ad platforms and ad networks in this QuantMar thread.
ATT, by reducing the measurability of advertising campaigns (especially given the acute deficiency of SKAdNetwork), exacerbates this two-tiered system of measurement: the largest ad platforms have the leverage to convince advertisers that their reporting should serve as a single source of truth. My sense is that this takes the form of ad platforms requesting that advertisers send all conversion data to them via API so that any given ad platform can probabilistically determine which of those conversions it delivered.
This is understandable, as currently, the largest ad platforms are having their conversions undercounted disproportionately relative to broker ad networks. This is because ad networks are currently able to use probabilistic attribution (more on this later) to attribute installs, whereas ad platforms rely on SKAdNetwork for conversion attribution. A tremendous amount of signal is lost with SKAdNetwork because of its privacy threshold and its timer system. Modeling solves this partially; building an all-encompassing measurement model that accommodates every traffic source is a much more robust solution that also privileges these large platforms. My belief is that tools will be rolled out by erstwhile self-attributing networks that demand all conversion signals from advertisers and report platform performance in that way. As advertisers shift measurement methodologies into macro-scope, top-down strategies like incrementality measurement and media mix modeling, ad platforms will meet that demand with their own proprietary tools.
Platform noncompliance with ATT
This prognosis is something of a corollary to the previous one, and it is already taking shape. Given that Apple has allowed what I call “probabilistic install attribution using device parameters” (PIAUDP) to take place in the open, without any intervention, then it makes sense that advertising networks and platforms will push the limits of the ATT guidelines until Apple clarifies and/or enforces ATT policy.
Why hasn’t Apple addressed the use of IP addresses and other device parameters for the purposes of attribution? There are a few reasons, and I discuss them in this podcast, this interview with Ben Thompson of Stratechery, and this article, but briefly:
- Apple’s definition of ‘fingerprinting’ is vague and unspecific;
- It would be akward and incredibly disruptive for Apple to police fingerprinting by ad tech platforms through the app review process;
- Fingeprinting (in the form of PIAUDP) does not allow for the persistent identification of a device, and so it is only really used for install attribution, not behavioral profiling, which is what Apple designed ATT to disrupt.
— Eric Seufert (@eric_seufert) December 9, 2021
When I speak with Apple employees, I’m told that fingerprinting isn’t being policed because ATT will evolve over a multi-year arc and that Apple’s broader privacy campaign (across initiatives like ITP, Private Relay, ATT, etc.) will ultimately revoke all access to a device’s IP address, rendering fingerprinting for in-app traffic mostly ineffective. In the meantime, however, I expect that the entirety of the ad tech ecosystem will push the limits of what can be accomplished through PIAUDP in order to regain measurement effectiveness against the pre-ATT standard. Reading between the lines with various product announcements from different platforms, my sense is that these systems are ready to be deployed. The Financial Times reported on an alleged program at Snap that joins conversion data to click data with device parameters through a partnership with attribution companies. I expect to see similar efforts become commonplace in the absence of reproach or enforcement from Apple over PIAUDP until IP addresses are fully obfuscated for in-app traffic.
Creeping temptation of brand-centric thinking
One barrier to advertising performance measurement resulting from ATT is the limited number of campaign identifiers that can be used on any given channel with SKAdNetwork. The consequence of this limitation is that creative testing is materially constrained, as I detail in Creative paralysis: ad creative production and testing in iOS 14. The prior strategy of mass-testing very many creative concepts, and permutations of each, is not workable with ATT: an app can only utilize 100 campaign identifiers per channel, and any combination of audience targeting and ad creative requires its own identifier in order to be measured. Creative testing is far slower and far less measurable after ATT relative to the previous, dominant approach I outlined in Mobile ad creative: how to produce and deploy advertising creative at scale.
This new reality may convince some advertisers that direct response advertising is less viable now than pure-play brand advertising — if direct response ad performance can’t be measured capably, why not lean fully into brand marketing? The temptation will be strong for teams without a core competency in analytics and marketing science to abandon direct response advertising completely, or in large part.
This is a mistake. I argue that brand marketing for digital products is mostly misguided in The perilous mythology of Brand Marketing for digital products, and I think teams should resist the urge to allocate budget away from direct response advertising and into brand marketing as a response to ATT. While investing heavily into brand marketing might superficially appear to be a strategic adjustment to ATT, it almost certainly won’t be: it’ll be an activity for the sake of activity as opposed to a complete adaptation to a new operating environment. The hard work of rebuilding new reporting and measurement infrastructure is the challenge that all advertisers must embrace.