App Tracking Transparency (hereafter, “ATT”) is in the news again because many advertising-supported companies have reported a particularly bad earnings quarter attributable, many of them have said, to several factors, perhaps best summarized by Mobile Dev Memo’s Eric Seufert:
It’s impractical, if not impossible, to try to tease out the individual burden of any of these dynamics on mobile advertising performance, generally. And it’s also largely beside the point: it is the “perfect storm” combination of these three conditions that compounds to such painful detriment to advertising performance.
This is perhaps true, but it has not stopped Seufert and others from calling out ATT as a key factor. Seufert published the third instalment of his series about how unfair ATT is earlier this month after news broke of new App Store ad formats, and it is, as is typical, an excoriation of the Apple-imposed question of whether users want to be tracked by third-party services:
Note that Apple’s ad network utilizes app install and in-app purchase data, to which Apple has exclusive first-party access under the restrictions of ATT, to target ads to users with its ad network. It’s worth underscoring that, with ATT, the scope and substance of consumer data utilized to target ads remains unchanged, except that only Apple has access to it. To be fair: Apple does employ privacy controls with its own ad network that are superior to the pre-ATT status quo. But my primary contention with ATT is that it does not facilitate real consumer choice and that it deprives consumers of widespread ad relevancy and advertisers and publishers of commercial opportunity.
Those are actually three “primary” concerns, and I think it is worth responding to them. But first, I think we should ask whether ATT really is cratering mobile advertising in the way both its critics and its proponents seem to believe. That includes me, by the way. I have previously linked to stories about the apparently enormous impact ATT has had on big ad companies like Alphabet, Meta, and Snap. But I thought it would be worth a deeper look.
As Seufert says, it is very difficult to figure out what specific effect ATT has because there are so many factors involved. But it is fair to think that, if it is affecting publishers’ revenue as Seufert says, it should also be affecting advertisers’ revenue too. And, while these companies do not separate revenue by platform, they do offer geographic breakdowns. North America is the only region where the iPhone is more popular than Android; elsewhere, the reverse is true, and often overwhelmingly. We also know ATT was rolled out at the end of April 2021. With time given for users to update, that means we should begin seeing North American revenue beginning to falter in the third calendar quarter of 2021 compared to the rest of the world.
The actual figures tell a much murkier story. I do not think it is fair to suggest ATT does nothing, but its effect does not seem as pronounced as either its biggest supporters or its biggest naysayers suggest.
Snap, for example, is a company that has no major revenue stream outside of ad placements in its smartphone apps. But in Q3 2021, a full quarter after ATT’s public debut, Snap posted year-over-year revenue growth of 57% overall. In North America, it reported 60% growth — higher than in any other region.
The following quarters all show overall revenue gains in North America just one percentage point below the company’s total growth. It is a pattern that more closely mimics the number of daily active users. Snap has only posted modest, single-digit year-over-year gains in North American users, but decent double-digit growth elsewhere. Meanwhile, its growth in the average revenue per user has been stronger in North America since ATT’s debut than anywhere else.
If ATT were so significantly kneecapping revenue, I would think we would see a pronounced skew against North America compared to elsewhere. But that is not the case. Revenue in North America is only slightly off compared to the company total, and it is increasing how much it earns per North American user compared to the rest of the world.
What about Alphabet? It has actually posted year-over-year revenue gains in the United States — one of few countries where iOS is dominant — higher than those in Africa, Asia, or Europe in its first and second quarters this year. In fairness, its gains were much stronger in “Other Americas”, which comprises Mexico, southward, plus Canada.
Meta’s business is the one everyone appears to be watching because two quarters this year have been rough. In its most recent, it reported its first ever year-over-year revenue decline, which dropped by about a billion dollars in Europe and about $600 million in the U.S. and Canada. That is alarming for the company, to be sure, but it still does not track with ATT causality for two reasons:
iOS is far more popular in the U.S. and Canada than it is in Europe, but Meta incurred a greater revenue decline — in absolute terms and, especially, in percentage terms — in Europe.
Meta was still posting year-over-year gains in both those regions until this most recent quarter, even though ATT rolled out over a year ago.
Those are all big, well-known companies. What about pure advertising businesses? Surprisingly few are publicly traded. Even so, I pulled the earnings from a few popular programmatic display ad providers. Magnite, for example, calls itself the “world’s largest independent sell-side ad platform”. In its most recent quarter, the proportion of revenue it derived from the U.S. increased year-over-year compared to the rest of the world. The most recent investor report from Criteo, a major provider of retargeted ads, showed an overall decline year-over-year, but the Americas performed far better than African, Asian, or European markets.
Perhaps the most favourable evidence for ATT’s effects lies in the earnings reports from Publicis Groupe, which has acquired dozens of name-brand agencies — like Leo Burnett and Saatchi & Saatchi — and also runs a digital ad platform. It is such a multifaceted business that it is hard to see where the effects of ATT may come into play. In the first half of 2022, its “organic” growth in North America was the lowest of any region. But it ranked among the middle in total growth over 2021, posting higher gains than Asia or Europe. In the same press release, Publicis also specifically calls out the performance of Epsilon, its internal data brokerage service, as a reason for its U.S. growth.
Though I did not examine every available earnings report, I am not cherry picking. I looked through the list of companies on Martech Map, checked to see if they were significant enough and had investor information, and then looked for geographic breakdowns. It is possible I have my assumptions all wrong, too; please let me know if you believe that is the case. I am not arguing this is a perfect analogue, only that it paints a murkier picture of ATT’s apparent financial effects on the ad tech industry.
I think Seufert’s criticisms of ATT have been among the most cogent and thoughtful, and I do not intend for this to be a full article about him, specifically. But he does articulate some of the more common problems I see being raised with ATT. There are legal questions which are being investigated by British and German authorities about Apple’s simultaneous offering of “personalized” App Store ads; I will focus only on the moral questions on which I think can fairly comment.
There is a fairly significant ethical problem out of the gate: there are those who believe highly-targeted advertisements are a fair trade-off because they offer businesses a more accurate means of finding their customers, and the behavioural data collected from all of us is valuable only in the aggregate. That is, as I understand it, the view of analysts like Seufert, Benedict Evans, and Ben Thompson. Frequent readers will not be surprised to know I disagree with this premise. Regardless of how many user agreements we sign and privacy policies we read, we cannot know the full extent of the data economy. Personal information about us is being collected, shared, combined, and repackaged. It may only be profitable in aggregate, but it is useful with finer granularity, so it is unsurprising that it is indefinitely warehoused in detail. You can prove this to yourself by viewing the browsing history collected by Facebook and Google, or requesting a copy of your personal data from major brokers. Some make that process very easy: you can often complete a form on the company’s website. Others require you to send an email with the personal identifiers you would like to obtain a records check on, like your name, email addresses, phone number, and device IDs. Some will display user data to those who ask anywhere in the U.S. or worldwide, while others will only comply with requests from California or Vermont, or wherever laws require. You may find some companies you have never heard of have a lot of information about you, often a mix of scraped public sources and data shared or collected in private deals.
What you will likely find after completing several of these requests, especially if you live in the U.S., is how much information about you is being held by by these brokers and marketing companies. Even though Canadian privacy laws give me some cover from the worst abuses, I have still found brokers that held my full name, my full street address, and other personal identifiers. These attributes are not often not relevant to targeted advertising — what does it matter what my apartment number is? Why are brokers not dividing the world into general areas in the style of What Three Words? — but they hold it all because it is cheap enough to do so, even at scale. All so, the story goes, a neighbourhood restaurant can precisely advertise a special offer when it is close to my partner’s birthday.
In a passionate defence of targeted ads, Seufert asked, rhetorically, “what happens when ads aren’t personalized?”, answering “digital ads resemble TV ads: jarring distractions from core content experience. Non-personalized is another way of saying irrelevant, or at best, randomly relevant.”
For what it is worth, a relevant ad has never serendipitously graced my screen, even before I took steps to avoid targeted advertising. My friends and family barely see well-targeted ads, either. Most often, they see the same ad — on every other webpage and in every app they use — for an online store they visited once, begging them to return, sometimes in French when their device is set to an English language setting. What is the solution to this — more data collection? That is absurd. Even at their absolute best, targeted ads are seen by viewers as creepy. People do not want irrelevant ads, but they do not want to feel followed or harassed either. Targeted advertising enables the latter. Even if they were a significant payoff for publishers — which there is not — does it make sense to build the internet’s economy on the backs of a few hundred brokers none of us have heard of, trading and merging our personal information in the hope of generating a slightly better click-through rate?
Earlier, I quoted Seufert:
But my primary contention with ATT is that it does not facilitate real consumer choice and that it deprives consumers of widespread ad relevancy and advertisers and publishers of commercial opportunity.
ATT may not be worded fairly — though Seufert’s proposed solution is similarly vague and unhelpful — but he is right to argue it does not offer real choice, though probably not in the way he intends. Users can still be tracked and apps from well-known developers were found to ignore opt-outs.
Then there is the much bigger question of whether people should even be able to opt into such widespread tracking. We simply cannot be informed consumers in every aspect of our lives, and we cannot foresee how this information will be used and abused in the full extent of time. It sounds boring, but what is so wrong with requiring data minimization at every turn, permitting only the most relevant personal data to be collected, and restricting the ability for this information to be shared or combined?
Does ATT really “[deprive] consumers of widespread ad relevancy and advertisers and publishers of commercial opportunity”? Even if it does — which I doubt — has that commercial opportunity really existed with meaningful consumer awareness and choice? Or is this entire market illegitimate, artificially inflated by our inability to avoid becoming its subjects?
I wonder how much of ad tech’s woes is really ascribable to ATT, and how much is the fault of the myriad other problems it is running into: currency fluctuations, regulation, pandemic effects, and changes in user behaviour all come to mind.
Cal Newport, the New Yorker:
[…] TikTok is estimated to have a billion active monthly users, a number it achieved in a breathtakingly short time, and according to some reports it boasts an average session length of 10.85 minutes, which, if true, would be far longer than that of any other major social-media app. Meanwhile, Facebook’s parent company recently lost more than two hundred and thirty billion dollars in market capitalization in a single day after the company announced that user growth had stalled. Analysts identified TikTok as an important factor in this slowdown.
Is it possible the social media giants from California are facing waning relevance? Is ATT perhaps a useful scapegoat with questionable effect? I am not sure it is possible to say from the outside looking in, but I am also not sure we can draw any conclusions from one or two quarters this year, over a year after ATT was launched to the public.
In theory, ATT is a very good option for users. Its biggest problem is that the company which makes it also has an advertising division, and it appears to have engaged in some quiet self-preferencing behaviours. Legal questions aside, it is disappointing to see such an obvious user benefit so easily undermined. These App Store ads give ATT’s critics a clear conflict of interest to point to, look tacky, and create an unpleasant experience. ATT’s reliance on a very specific definition of “tracking” that allows Apple to segment users based on what they read in News and what they buy in third-party apps is far more permissive than I think it ought to be for a company that so loudly trumpets its privacy bonafides. But advertising that relies on first-party data can accurately be described as better for privacy than those based on the third-party data economy. Whether it is fair for Apple to treat itself, as the platform creator, as the root-level first-party with an infinitely bigger observation window is another question. I do not think it should.
Conflicts like these are one of many reasons why privacy rights should be established by regulators, not individual companies. Privacy must not be a luxury good, or something you opt into, and it should not be a radical position to say so. We all value different degrees of privacy, but it should not be possible for businesses to be built on whether we have rights at all. The digital economy should not be built on such rickety and obviously flawed foundations.