At SMX Advanced in Boston earlier this month, I attended back-to-back sessions of Crystal Carterhead of AI research and SEO communications at Wix, and Jen CornwellSenior Director of AI SEO at Tinuiti. On paper, they covered the same pace: how AI research is reshaping the marketer’s work. In the room, they couldn’t have approached things more differently. This gap turned out to be the most useful thing the conversation taught me.
After the conference, I emailed Crystal and Jen and received a copy of both presentations to make sure I represented what they said at the event.
Carter’s Talk: A Framework for What to Optimize
Carter’s session rests on a distinction that does most of the conceptual work. Memory is what an AI assistant passively infers from the way you speak to it, from your tone, from your complaints, from your patterns. Personalization is what you actively declare, via profile settings, connected applications and declared preferencesand it has enough weight to shape what an agent actually does, not just what it appears to do. You can’t work your way into someone’s inferred memory the same way you would adjust a meta-description, but you can design the cues that shape both halves at once.
The most concrete evidence she provided wasn’t a best practices slide. This was an iPullRank experiment using three accounts running identical prompts with different levels of personal data connected, which produced visibly different responses in AI modeincluding a response that addressed a hypothetical child by name in a streaming recommendation. This is a controlled comparison, not an anecdote, and it’s the kind of detail that should worry anyone who still treats AI search results as a single, generic output that everyone receives the same way.
From there, Carter turned to tactics, starting with denominal nouns (“actor” instead of “the person who acted”) because semantic patterns cluster identity-related queries that way. And the average Google query contains three to four words, while the average ChatGPT opening prompt contains around 103 words. This gap is the argument in favor Narrowly specific, FAQ-style content on large landing pages. Users typing into an AI assistant are already further down the funnel than a search box ever got them.
Cornwell’s speech: a framework for why no one acts on it
Cornwell’s session contained almost no new SEO data, and that’s the point. She started by naming a completely different problem. Most research teams have no shortage of information; They lack an organization ready to act on the basis of the knowledge it already has. This is not a research problem. It’s a change management problem, and she handed the room two borrowed frameworks to solve it, Kotter’s Eight-Step Model of Change And Everett Rogers’ Diffusion of Innovation Curve.
The device she used to make it stick was Kotter’s own 2005 fable about a colony of penguins on a melting iceberg, revamped with glimpses of AI like melting ice and five roles (Sponsor, Trust, Catalyst, Analyst, Skeptic) that each participant was implicitly asked to assign within their own team. By the closing slide, you were no longer taking notes on an eight-step process; you organized a casting on your own organizational chart.
The research anchor worth keeping is the Rogers tipping point calculation. Innovators make up 2.5% of the population, early adopters 13.5%, and once a change hits that 16% total, adoption tends to become self-sustaining. Applied internally, this rephrases “convince the entire company” to “find the findable minority,” which is a much less paralyzing target for an SEO arguing for budget to a room full of skeptics.
Where the two talks actually collide
Here’s the dissonance and why it’s worth more than speaking alone. Carter’s framework assumes that the bottleneck is knowing what to build, the right structured datathe right niche content, the right MCP server setup. Cornwell’s framework assumes you already know what to build, and the bottleneck is ask five other departments to let you ship it. Put them in the same room and they stop feeling like two lectures on the same topic. They’re starting to feel like a diagnosis of why so many AI research initiatives are stagnating; most teams only have tools for half the problem.
If your AI research strategy has a technical roadmap but no internal coalition, Carter’s tactics will remain in a package that no one will approve of. If you have management buy-in but no specific plays to execute, Cornwell’s framework will produce a motivated team with nothing concrete to do Monday morning.
3 movements to perform in both rooms
- Choose a niche content gap, not a full audit. Use Carter-owned channel framing, but resist creating a comprehensive AI visibility document that no one reads. Ship an FAQ-style piece of content that matches how people actually invite AI assistants, then use it as a proof of concept internally.
- Find your 16% before presenting the whole room. Identify one or two people already convinced about investing in AI research and develop your brief with them first. You don’t try to convince your most skeptical stakeholder on day one.
- Choose your five roles before the next proposal. Name who on your team is the sponsor, the skeptic, the catalyst. Going into a budget conversation already knowing where the resistance will come from is worth more than another slide of AI mode screenshots.
Put Carter and Cornwell next to each other and it’s hard to miss the lesson. Most teams view AI research as two separate tasks: those who figure out what to build and those who fight to get it shipped. Carter’s room assumed the hardest part was knowing what to optimize. Cornwell assumed you already knew that, and the real work was getting everyone to act on it. Both are right, that’s exactly the problem.
A technical roadmap without an internal coalition stagnates in a game that no one approves of. A motivated team without a specific game has nothing to do on Monday. The strategies that really move are those that are executed as one task, not two. Optimization has never been the hard part. Getting your organization to act accordingly is.
More resources:
Featured image: Master1305/Shutterstock
