Our friends at Future Commerce recently conducted an interesting two-phase holiday shopping study in partnership with Cimulate.

The main takeaway is that, at least for some consumers, AI platforms are now the front door to commerce. 80% of AI-native consumers surveyed began their gift research on these platforms—not on Google, not on Amazon, and not on brand sites.

At the same time, AI isn't stealing conversions from DTC (yet!). 77% of shoppers who receive a product recommendation from an AI platform still click through to the merchant site to research or complete their purchase. So traffic is still coming, but shoppers are arriving with radically different expectations.

What are those expectations exactly, you ask? Frankly, it’s too early to tell: as AI platforms mature and continue to improve their native checkout experiences, they will capture a larger share of transactions. Still, our hope is that agentic shopping will free up DTC to do what it does best: tell your brand’s story in unique ways and build loyalty through exclusive experiences.

OpenAI used to say they'd never do ads. Now OpenAI says that okay, maybe they need ads, but the ads won't affect your ChatGPT answers. Meanwhile, Anthropic is running a funny if somewhat obnoxious campaign promising Claude will stay ad-free.

This holier-than-thou battle about AI ads is naive and, frankly, it's getting boring already.

If you thought AI would never get ads, you haven't been paying attention to the last two decades of Internet history. Every new channel promises to help users cut through the noise. Every new channel inevitably rolls out ads and ends up contributing to said noise. Rinse and repeat. The economics have never worked any other way, and they certainly won't start working another way with AI.

The good news is that, ultimately, no one cares whether ChatGPT shows you ads, or whether the ads affect your answers. What everyone cares about is that (1) user privacy is respected, (2) ads are genuinely useful, and (3) platforms offer an escape hatch in the form of a paid plan for professional/corporate use cases.

Over $227 million has flooded into just a handful of startups promising to optimize your brand for ChatGPT. What do they have to show for it?

In a recent article, Scot Wingo at Retailgentic (highly recommended) broke down the AEO/GEO space, trying to separate signal from noise. His take—and we agree—is that the AI visibility metrics these tools trumpet are close to useless.

Part of the problem is the data itself. Most tools either rely on AI-generated prompts or source "panel data" from browser extensions that track every ChatGPT conversation users have. This is one data breach away from a PR catastrophe.

But even if the data were first-party and ethically sourced, the results would still not be repeatable. LLMs are non-deterministic and personalized: ask ChatGPT "what are the top-rated sneakers" from two accounts and you'll get different brand mentions.

What's actually useful: crawlability improvements, clean structured data, and brand/product storytelling—all things you should be doing anyway. The rest is theater until AI platforms release real analytics—which, by the way, Bing just did.

💸 Panic discounting is liquidating your brand. In our latest blog post, Andrea breaks down how brands get stuck training customers to wait for sales, and walks through a recovery roadmap that doesn’t tank your revenue.

🛠️ We're hiring Shopify engineers (again!). We're looking for someone with the consulting and technical chops to shape, build, and ship e-commerce systems transacting hundreds of millions in GMV. Is that you?