I wrote on Tuesday that creative production is a poor choice as an initial application of AI tools to a marketing workflow. Which raises a question: what's a better option? Where would AI enablement produce more value? I think there are three (product) marketing activities where

Netflix’s YouTube opportunity Netflix revealed in its engagement report for 1H 2025 that the first season of "Ms. Rachel," the series it contracted from YouTube creator Rachel Accurso, was the seventh most viewed program on the platform. Notably, Season 1 consisted of only four

@antoniogm As @MobileDevMemo has said, *of course* it’s gonna be ads. Just need the doomers and boomers to let the practicals get a word in

Agentic commerce is a mirage If agentic commerce becomes a well-worn consumer behavior, why would it not be catalyzed and captured by the largest retail platforms themselves? The digital advertising ecosystem is instructive: the open web has given way to walled gardens and

Seems like Google Search remedies are a nothingburger.

Google’s Gambit and the future of the open web I’ve described Google’s ambitions with AI Overviews and AI Mode as Google’s Gambit: an attempt to utterly reform the core Search experience through AI functionality while not alienating users. The Gambit appears to be paying off.

I’ve seen the practices of large companies be characterized definitively as violations of Apple’s AppTrackingTransparency (ATT) privacy policy. Most recently, this assertion surfaced in relation to Meta’s 2023 overhaul of its AEM attribution framework in a lawsuit against it from

In 2021, I wrote “The Growth Trap,” in which I described a tendency for consumer product companies to optimize their products for the early cohorts of users that discovered them organically, ignoring the preferences and usage patterns of users acquired with ads, which might

The application of AI to marketing is transformative. Counterintuitively, many teams are devoting too much effort to it: - Implementations of AI tend to focus exclusively on the outputs of these tools. Most often this means creative production. But much of this functionality

When I discuss "signal engineering," I'm often asked if I've written any articles on the topic. As a brief primer: signal engineering is the process of designing and capturing high-quality outcome proxies (eg., pLTV) so they can be reported back to ad platforms in real time

How AI efficiency could shape ad load Ad sequencing becomes a much more important consideration in the ad serving process as advertising is optimized with AI: if a consumer’s disposable income is depleted, the conversion probability for all future ad exposures falls sharply. As

Snap Q2 2025 earnings: eCPM decline, DR revenue slowdown https://t.co/NMtGsFLzWf

This is the fruit of Meta’s strategy to navigate the restrictions of ATT: increased engagement through open-graph short-form video to compensate for a lack of device identifiers. I discussed the motivation in this piece from July 2022:

The bifurcation of outcomes in Q2 earnings for advertising platforms reveals the degree to which value is derived from data in an environment where AI is applied to every step in the advertising value chain. Every platform oversees a large volume of data. But not all of that

In this week's episode of the @MobileDevMemo podcast, I speak with @aepstein_ about Amazon's advertising strategy. Among other things, we discuss: The partnerships that Amazon announced with Roku and Disney at Cannes Why Amazon’s dataset is a natural fit for CTV, and whether

AppLovin Q2 2025 earnings: ad revenue up 77%, general self-serve in 2026 Advertising revenue grew by 77% to $1.26BN. AppLovin's CEO noted in prepared remarks that the company’s self-serve platform for eCommerce advertisers would expand beyond US advertisers in October and be

AI is not the Metaverse A pervasive narrative frames large-scale investments in AI by the behemoths of digital advertising as speculative, long-term bets with potential payoffs that invite scrutiny. But this view ignores the fact that AI investments have already yielded

Last week, AI researchers at Meta published a paper that finds its LLM for call-to-action text, AdLlama, improves advertising CTRs by 6.7% when post-trained using click data. Post-training refers to the stage in LLM development that adapts a pretrained model to a specific use

$META's investments into AI infrastructure (Generative AI and "Core AI") should be understood in the context of its principal customer base, which is SMBs whose budgets are tightly tethered to performance: - improvements to Andromeda and GEM may have "only" driven single-digit

The first review of my book Yield is out, from none other than @eric_seufert in his excellent @MobileDevMemo newsletter. He has some nice things to say! Link below 👇