From Amanda
Netflix Just Pulled a Page From the Performance Playbook
Netflix announced an expansion of its Ads Suite, including the launch of its own Conversion API (CAPI).
When Netflix launched its ad-supported tier in 2022, the pitch was reach and premium inventory. But advertiser expectations have shifted. Marketers don't just want reach anymore; they want proof that ads drive outcomes. Plus, Netflix's ad revenue crossed $1.5B in 2025, and it set a $3B target for 2026. Hitting that number means convincing performance advertisers that Netflix can deliver the targeting and measurement they expect elsewhere.
CAPIs are already table stakes on platforms like Meta, Google, and TikTok, but they're newer to streaming environments. The rising pressure to justify TV ad buys is exactly why streaming companies are now building this infrastructure.
Instead of relying on browser pixels, which Netflix doesn’t offer, a CAPI sends data server-to-server — letting advertisers share conversion events and campaign data directly with an ad platform, with cleaner signals and better resilience in privacy-constrained environments. Netflix says early tests with Tinuiti outperformed benchmarks by more than 75% across financial services, ed tech, and retail.
Netflix also announced integrations with Amazon DSP and Yahoo DSP. The Amazon partnership is worth pausing on. Netflix and Amazon first announced their demand partnership in September 2025, but that integration didn't initially include Amazon Audiences — the part that matters most for performance advertisers, since it lets advertisers layer Amazon's shopping, browsing, and streaming signals onto Netflix inventory.
That's notable because Amazon is a direct competitor in streaming, running its own ad-supported tier, its own measurement infrastructure, and its own demand-side platform (DSP). It's a sign of how data partnerships now cut across competitive lines because everyone wins enough to make it worthwhile. As Amazon Ads’ Sarah Iooss put it, the goal is to “improve ad performance by connecting shopping and streaming data.”
With commerce signals plus streaming inventory plus server-side measurement, Netflix is starting to look less like a TV network and more like a modern performance platform. And once a platform can connect ad exposure to outcomes, it stops being just a brand channel and starts competing for performance budgets.
Tech Giants
Android Tweaks the Play Store as Pressure Builds on the “App Store Tax”
Great news for developers on Android — Google is cutting Play Store fees and offering more billing flexibility. But Google didn’t do this out of the goodness of its heart.
Google’s legal battle with Epic Games pushed it here. This might sound familiar, because Apple already made changes based on similar rulings last year — but Google's outcome was far more consequential, and the changes to the Play Store reflect it.
Google is now decoupling the billing fee from the service fee entirely. Developers who use Google Play's billing system will now pay a separate 5% billing rate, with service fees dropping to 20% for new installs — and as low as 15% for developers participating in certain Google programs. The service fees for recurring subscriptions fall to 10%.
Developers will also be able to offer alternative billing systems alongside Google's own or direct users to their websites for purchases — a more flexible setup than Apple’s current model, which mostly allows developers to link out to external payment options.
Google is also introducing a Registered App Stores program — an optional, streamlined installation flow for third-party stores that meet Google's quality and safety requirements. While Android technically allowed sideloading before, Google’s warning screens and installation friction made it difficult for rival stores to reach mainstream users. This lowers those barriers in a meaningful way.
That said, the rollout is gradual — the EEA, U.K., and U.S. see changes by June, with global expansion running through September 2027. And for most mobile marketers, nothing changes tomorrow. It’s worth making sure your attribution can follow a journey that exits the store — otherwise you’ll lose visibility right where the complexity is growing.
AI
People Are Ditching ChatGPT for Claude
After public debate over how AI companies handle government access, privacy, and ad monetization, users are paying much closer attention to who they trust with their data. And they're moving. Claude recently hit No. 1 in Apple's U.S. App Store free rankings, blowing past ChatGPT — and tutorials on how to migrate chat history from ChatGPT to Claude started circulating almost immediately.
That says a lot about how people use these tools now. They're not just utilities. They're places where people think out loud, solve work problems, and build context over time. So when something changes — ads appear, policies shift, or the product starts to feel off — switching feels risky, because your history with the tool feels irreplaceable.
But how much does your history with the chatbot actually matter? LLMs do not actively think through your entire archive every time they answer. Every model’s context window — the model's working memory — is finite, and once a conversation exceeds that window, it has to be truncated or summarized. OpenAI's own memory documentation indicates that chat history is used selectively, saved memory is meant for high-level preferences, and the system does not retain every detail from past chats equally.
There's a practical reason for this. Longer context windows cost considerably more — compute requirements rise sharply with context length. So natural economics push AI systems toward using the smallest amount of relevant information needed: recent context, compressed summaries, and selectively retrieved memory. This is why even the newest frontier models often default to summarized context or smaller helper models — it costs less and uses less memory.
Which means the stickiest and most valuable thing you’ve built with any AI tool probably isn’t stored on their servers. It’s the instinct you've developed for how and when to use it — how to prompt, what to challenge, how to frame the problem, and how to recognize a good answer when you see one. And familiarity is a lot more portable than it feels.
Advertising
Apple Is Adding More Ad Placements to the App Store
March is here, and that means Apple is delivering on its teased plan to expand ad placements in App Store search results. Sponsored listings won’t appear just at the very top of results anymore — they can also appear further down the page.
The rollout is starting in markets like the U.K. and Japan before expanding globally by the end of March.
For years, the search results page had a relatively stable structure: one sponsored result at the top, followed by organic listings. Now additional ad placements mean paid listings may appear deeper in the results, potentially pushing organic results further down the page.
More ad slots create more opportunities to win installs. But they also mean more competition on valuable keywords and potentially higher costs for install-driven campaigns.
It’s also worth noting that, contrary to the many hopeful comments I’ve seen wondering if these placements would come at a discount, billing remains unchanged, with advertisers paying on a cost-per-tap or cost-per-install basis.
Adam's Take
OpenAI Scales Back Shopping Plans for ChatGPT — but Agentic Commerce Isn’t Dead
Reports that OpenAI is scaling back its agentic commerce plans have AI-shopping bears crowing the whole concept is dead — mainly because it violates retailer motivations.
It’s a fair point: OpenAI promised a lot and seems to be throwing up its hands after five months with only a dozen Shopify merchants to show for its efforts, but I don’t think this means we’ve seen the end of agentic commerce. Sure, it doesn’t look like we’ll be asking AI to select and buy our summer wardrobe, but LLMs are already changing product discovery for a majority of shoppers — and changing discovery changes shopping behavior.
Retailers dragged their feet on e-commerce too. Who can blame them: Why risk loyal brick-and-mortar customers or their powerful, proven advertising pipelines? In the end, it didn’t matter. Consumers adopted a new technology, and the retailers that didn’t lean into online shopping were dragged into it kicking and screaming.
Agentic commerce is already a thing. Today it’s helping with discovery and product recommendations, tomorrow it’ll be an automated checkout process (goodbye filling in shipping and credit card information), and next year it’ll auto-order eggs or other regular consumables. Eventually consumers will be ready for instant checkout on some items. But like e-commerce, it’ll be a slow process of adoption. And if current retailers don’t like it, there'll be plenty of new ones ready to fill my automated order.
Has Mobile Gotten Boring?
Last week Apple launched new phones — or was it the week before? Seminal Apple journalist John Gruber summed up the launch in an uncharacteristically dry update. My takeaway on his commentary: “That’s about it.”
Has mobile gotten dull? Or is it that mobile news is being usurped by gestures broadly at everything?
The truth is, 481 days short of its 20th anniversary, the smartphone has gotten pretty boring. It’s a great device, it gets better every year, everyone has one, and … life goes on. That’s not to say the industry is shrinking: It’s just that device sales are mostly replacements, and for the first time, games revenue serves as the minority in the market.
What we’re seeing is the maturity of a market. Mobile is no longer a land-grab game; we’re moving from a market driven by adoption to a market driven by optimization. We used to refer to mobile apps as the “wild west” — now that’s a title held by AI (or more recently, the Middle East). The gaming-heavy mobile economy rewarded creativity. A two-person start-up could throw a binary into the App Store over the weekend and — with a little luck — get a viral hit. Now the economy rewards subscription mechanics, cross-platform retention mechanisms, enterprise-like operating discipline, and large installed audiences. In other words, winning has changed from “go viral” to “build a durable business.” That’s not nearly as interesting. We used to get company updates from the big mobile CEOs on Twitter or on a small conference stage. Now we’re hearing CFOs read carefully worded announcements on quarterly earnings calls.
So yes, Apple phone releases aren’t front-page news anymore. But that’s not because mobile is dying. It’s because mobile has made it. The smartphone is no longer a novelty; it’s infrastructure. And infrastructure is rarely exciting. The Industrial Revolution may seem boring, but what it produced was extraordinary: industries large enough, durable enough, and important enough to shape how we live, work, shop, and play. Mobile may be boring, but that’s because it’s grown up.
Podcast
AI won’t replace your artists, but it will free them to create more with Jen Taylor
In this episode of How I Grew This, host Amanda sits down with Jen Taylor, Director of AI Strategy and Integration at Capacity Interactive, alongside co-host Adam, to explore how arts and culture organizations can adopt AI thoughtfully, ethically, and strategically. Drawing from her 15 years building audiences across streaming and entertainment, Jen shares a practical roadmap for bringing AI into creative industries while keeping human expression at the center.