The question many B2B marketers seem to be asking about AI is: “how do I show up in ChatGPT?” That’s a reasonable question. It’s also the wrong starting point.
The right starting point is understanding what AI systems are actually doing when they generate an answer, and who they think they’re talking to when they do it.
AI doesn’t just retrieve “the answer”. It reasons.
When someone asks an AI a question, it doesn’t return a list of pages ranked by authority (a world we have all been used to for 20 years). It synthesises a response from what it’s absorbed, weighted by what it trusts, shaped by the context of the query. That’s a fundamentally different process from search.
The implications for B2B marketers are significant. Generative Engine Optimisation isn’t about keyword density or backlink profiles. It’s about becoming the source a model reaches for when constructing an answer about your category.
To do that, you need to be legible to the model. Not just indexed. Legible. Clear, consistent, authoritative, cited by others, and present across the places models draw from: forums, publications, academic papers, structured data, third-party platforms.
Most B2B content is none of those things :). It’s SEO-optimised landing pages and gated whitepapers that models can’t read and wouldn’t trust if they could.
The brand problem hiding inside the AI problem
Here’s what makes this genuinely hard: AI visibility is a downstream consequence of brand. If your category doesn’t associate your company with solving a particular problem, a model won’t either. Models are, in a crude sense, a compressed representation of what the internet believes. If the internet doesn’t believe you’re the authority on something, the model won’t say you are.
This is why I think the “GEO as a tactic” framing is mostly wrong. You can’t optimise your way to AI visibility if you haven’t done the work of being genuinely known for something. The original GEO research from Aggarwal et al. at Princeton, Georgia Tech, and the Allen Institute for AI (arXiv:2311.09735) looked at citation and fluency strategies, but even their highest-performing interventions depended on content that was substantive and specific, not generic optimisation.
The tactics matter. But they amplify signal. They can’t manufacture it.
What B2B marketers are actually missing
The practical gap I’ve most seen isn’t technical – it’s a measurement and priority problem.
Most B2B marketing teams are measuring AI visibility the same way they measure search rankings: spot-check a few queries, see if you appear, move on. That’s not enough. AI responses are query-specific, model-specific, and change without warning. You need systematic monitoring across the queries that matter to your buyers, not a monthly manual check.
The second gap is content architecture. B2B marketers have spent fifteen years building content for humans scanning a SERP. AI doesn’t scan. It reads. Long-form, specific, well-cited content that actually explains how something works tends to surface more reliably than thin pages optimised for a single keyword. Botify’s 2025 research on AI crawler behaviour is useful here, showing that models disproportionately pull from pages with higher word counts and more external references.
The third gap is attribution. If a prospect read a ChatGPT answer that mentioned your company, visited your site three days later, and converted via a paid ad, your last-click model says the ad worked. That’s why marketing measurement in an AI-mediated world is a problem worth solving properly, not patching.
AI visibility isn’t a new channel to add to the mix. It’s a lens that exposes how well your existing marketing is actually working.