The Layer Nobody Is Contesting

Every tactic that could be copied in a weekend has been copied, tested and, in several cases, found not to work. What remains is the product: what it is called, how it documents itself, whether its facts agree, and how much of it a machine can verify. That is where the next phase of AI visibility will be decided.

This book made one argument from several directions.

AI engines judge entities, not pages, and most of what defines your entity is produced by the product. The signals that matter are domain-specific, so the work is vertical. The engines are noisy, so the method is experimental. The product layer has four parts plus a baseline: naming, documentation, a single source of product truth with consistent self-description, and capability exposure, all measured against what engines actually say. And the work needs owners in product, connected by product marketing, running a steady practice rather than a campaign.

Why the window is open

Right now the GEO market is crowded in three of the four layers. Agencies, platforms and in-house teams compete over content, authority and technical signals every day. Very few are working on feature names, docs coverage, product fact models or agent-ready capability catalogs, because that work sits in the gap between the team that builds the product and the team asked to make it visible.

That gap is the opportunity, and it will not stay open forever. As the turf war settles and Google's "still SEO" message pushes teams away from tricks, more companies will discover that the remaining work is product work. The ones who start now will have cleaner product signals, longer experiment logs and better-calibrated practices by the time the rest arrive.

This is my analysis, not a forecast with a date on it. But every trend in the 2026 data points the same way: buyers shortlisting in AI, engines converging on brands, single tactics failing controlled tests, and agents starting to read products directly.

What changes on Monday

If you take one thing from this book into next week, make it the description test. Write the truth sheet. Ask the engines your buyers use to describe your product, many times each. Mark what is missing, wrong and vague. Then trace each problem to the layer that caused it.

You will probably find that most of the problems are not content problems. That finding, put in front of your product leadership with evidence, is how a product-resident GEO practice begins.

Back to the inversion

The book opened with falling impressions and rising AI citations on two properties I run. The lesson I draw from that inversion is not that traffic no longer matters. It is that the place where buyers form their view of your product has moved, and the material they read there is increasingly your product's own self-description, filtered through engines you do not control.

You cannot control the engines. You can control what your product says about itself, how clearly, how consistently and how completely. That has always been good product practice. In AI search, it is also how you get chosen.

One last question, the one I would ask every B2B product team: when a buyer asks an AI engine about your category tomorrow, will it describe the product you actually built?

Key takeaways

• AI engines judge entities, and most of what defines your entity is produced by the product.
• Three of the four layers are crowded. The product layer is open because it sits between the team that builds and the team asked to make it visible.
• Start Monday with the description test, then trace every problem to the layer that caused it.