Research

Search Is Moving Beyond Keywords

A working hypothesis on what happens when interpretation, intent and evidence become part of discovery.

Our Conclusion, Not Google's

This is our hypothesis. Below it, we show exactly the evidence it's built on.

Two systems are moving in a related direction, in parallel, without merging. On the organic side, AI Overviews and AI Mode add an interpretation layer between a question and the pages that answer it. On the paid side, AI Max adds an optimization layer inside Search campaigns. Neither replaces the keyword. Neither is the other system wearing a different label — and we're careful to keep that distinction sharp in the full paper.

The single clearest piece of evidence we found is Final URL Expansion: a marketer sets a destination page, and Google's system can send the click somewhere else instead, if it predicts better relevance. That's not us speculating — it's a documented Google Ads setting, with real constraints and one asset type it explicitly overrides.

What's Inside
AI Search ≠ AI Max. Why treating AI Overviews and AI Max as "the same shift" is a common but inaccurate reading — and what actually connects two systems that share a direction without sharing an architecture.
AI Max, precisely defined. Google's own words: a continuous optimization layer for Search campaigns, not a separate campaign type — bundling search term matching, text customization and Final URL expansion.
Final URL Expansion, in full. The relevance constraint, URL exclusions and inclusions, why tracking templates still apply, and the one control — pinned RSA assets — that gets overridden. The clearest concrete example of the whole thesis.
A different unit of discovery. Our own model, labeled as such: Intent × Entity × Context × Evidence — and why it's a hypothesis, not a Google framework.
Three evidence tiers, marked throughout. Every claim is tagged Google Documented, Observed, or AdsWizards Lab Hypothesis — so you can see exactly how much weight each one carries.
The Experiment

AdsWizards Lab is part business, part experiment. We deliberately separated conventional search optimization from AI visibility when building this site — building a connected knowledge structure instead of manufacturing keyword-targeted pages, and attempting initial visibility without paid acquisition. We don't yet know if the hypothesis holds. We'll publish what we learn, including the parts that fail. The full research walks through the evidence and the open questions.

← Back to Research library