THE 30-SECOND TAKE

ChatGPT launched advertising this week at $60 CPM with no conversion data. Read that again. Advertisers are paying 3x Meta rates for reach they can’t measure. This isn’t a temporary gap, it’s the new reality. The measurement system that built modern marketing assumed humans visited websites we could instrument. AI agents don’t.

Why this matters for executives: Your board will ask about AI advertising spend within 6 months. You need a position on whether to pay premium for unmeasurable reach, and what that means for your measurement philosophy overall.

Why this matters for digital practitioners: The attribution models you’ve built your career on are breaking. This isn’t a temporary API gap. Start building expertise in statistical modelling and econometrics now, because that’s what replaces pixel-based attribution.

KEY DEVELOPMENTS

The investment frenzy reaches escape velocity

Amazon committed $50B to OpenAI investment alongside their $8B Anthropic stake. Microsoft’s Maia 200 AI chip signals serious vertical integration. Platform companies aren’t just adopting AI, they’re becoming AI infrastructure. I think we’re watching the formation of something like Skynet, but with better PR. The anti-trust regulators won’t understand how to stop this because the market is locking in before they’ve finished their morning coffee.

Why this matters for executives: Your technology stack assumptions are about to get disrupted. If your cloud provider is also your AI provider is also your commerce platform, vendor dependency takes on a different meaning. Start mapping where you have single points of failure.

Why this matters for digital practitioners: The tools you’ll use in 18 months will come from these consolidated platforms. Get familiar with multiple ecosystems now. Lock-in is coming, and you want to choose it rather than have it chosen for you.

YouTube becomes the citation king

YouTube is now overtaking Reddit as the dominant citation source in AI responses. The reason is structural, not editorial. YouTube transcripts are timestamped, sequential, clearly structured. Reddit threads waste tokens on meta-discussion and tangents. When AI systems allocate roughly 2,000 tokens to top results, information density wins. My hunch is this creates an interesting arbitrage opportunity for anyone who can reverse-engineer what YouTube content structures (tutorials, explainers, Q&A) get cited most.

Why this matters for executives: If your brand doesn’t have a video content strategy, you’re becoming invisible to AI discovery. This isn’t about YouTube views anymore. It’s about whether AI systems can find and cite your expertise.

Why this matters for digital practitioners: Your AEO strategy needs a video component. Specifically, you need to understand which YouTube content structures (tutorial, explainer, Q&A) get cited in your vertical. Run the audits.

Agentic commerce infrastructure forms

Moltbot (aka Clawdbot, and then OpenClaw), a 24/7 AI agent with full system access orchestrated via Telegram, went viral. Kimi K2.5 open-sourced a 1T-parameter model with 100-agent swarm coordination. One developer built a functional web browser in three days with a single AI agent. The barrier to building sophisticated agent systems just dropped dramatically. What was cutting-edge six months ago is now freely available code. I’m not sure the market has fully processed what this means.

Why this matters for executives: Your customers will delegate purchasing decisions to AI agents within 24 months. Your website, your checkout flow, your customer service, all of it needs to work for machines, not just humans. This is infrastructure investment territory.

Why this matters for digital practitioners: Stop thinking about users. Start thinking about agents. The systems you build need to be machine-readable, machine-navigable, and machine-trustworthy. That’s a different design brief entirely.

Attribution systems enter collapse

Research on 12,855 social posts showed invisible engagement (save, search, drive-by) at 10.1% shop engagement versus 6.8% for visible metrics. The metrics most marketing teams optimise for don’t predict what actually matters. Meanwhile, ChatGPT’s no-attribution-data advertising launch signals that the platforms aren’t even pretending they’ll share measurement anymore. The infrastructure we built attribution on is fundamentally breaking.

Why this matters for executives: Your marketing measurement framework needs rebuilding. The CFO’s question “what’s the ROI of this spend” is about to get much harder to answer with the tools you currently have. Budget for new measurement approaches.

Why this matters for digital practitioners: Visible engagement metrics (likes, comments, shares) are poor predictors of purchase. Invisible engagement (save, search, drive-by) predicts better. Rethink what you’re optimising for, and start building measurement systems that capture what actually matters.

THE BIG QUESTION

When AI systems both generate the discovery, intermediate the transaction, AND control the measurement, who captures the value?

Here’s how I read it: we’re witnessing the hotels.com moment of the entire internet. Your website just dropped from being the paper to becoming a reference in the AI’s paper. Not only do you want to get referenced, you want to be the main paper that the AI’s rewrite is built on.

The brands that win will need to do two things simultaneously. First, build trust directly with humans through channels that bypass AI intermediation (apps, email, SMS, community). Second, optimise relentlessly for machines (structured data, semantic triples, Q&A format).

The CDP and attribution expertise that built modern marketing isn’t obsolete. But the substrate is shifting from human journeys to AI context. The discipline is the same, structuring data so it informs decisions. The consumer of that data has changed.

I think we’re in the schema stuffing phase of AEO, similar to the keyword stuffing era of early SEO. There’s probably a 12-24 month arbitrage window where structured content gives disproportionate advantage. Eventually AI will discount obvious optimisation. The smart play is to structure content that’s genuinely useful to both humans and machines, so when the algorithm matures, you still rank because the quality is real.

Why this matters for executives: You need a dual-track strategy: human trust and machine optimisation. Neither alone is sufficient. The brands that win will excel at both simultaneously. This requires organisational change, not just tactical adjustments.

Why this matters for digital practitioners: There’s a 12-24 month arbitrage window where structured content gives disproportionate advantage. Learn schema markup, semantic triples, and Q&A formatting now. First-mover advantage is real, but the window is closing.

YOUR MONDAY MOVE

Pick one query in your expertise area and run it through ChatGPT and Perplexity. Note what gets cited. Then ask the follow-up questions the AI naturally suggests. Document whether your content answers those follow-ups or whether competitors appear. AEO isn’t about winning the first citation in the conversation anymore. It’s about owning the entire question sequence.

Why this matters for executives: This takes 15 minutes and will show you exactly how visible (or invisible) your brand is to AI discovery. Do it yourself before asking your team to report on it.

Why this matters for digital practitioners: This is the new competitive audit. Make it a weekly habit. Track changes over time. The brands winning this race are the ones who are measuring it.

QUOTABLE TAKE

“Your website just dropped from being the paper to becoming a reference in the AI’s paper. Not only do you want to get referenced, you want to be the main source the AI’s rewrite is built on.”

---

This briefing is part of my weekly AI intelligence analysis. I've been tracking these patterns for 30+ weeks. If you're navigating the shift from search engines to answer engines, let's talk.