Most Marketing Strategies Were Built for a Search Engine That's Already Changing. Here's the Gap Nobody's Named Yet.
By
Izzy Gregorio
·
6 minute read
Strategy · AI Visibility
Most Marketing Strategies Were Built for a Search Engine That Is Already Changing
The plan driving your content calendar, your search priorities, and your website roadmap was built for an environment that is measurably changing underneath it. Not someday. Now, with data to prove it.
By Izzy Gregorio · Updated August 2026 · 9 min read
In short
Google ranks pages. AI recommends entities. Those are two systems with two definitions of visibility, and most strategies are built entirely around the first. The gap is not a content problem, it is an infrastructure problem: entity clarity, structured data, and citable content have to exist before volume does anything. Six dimensions determine whether you get named, and most plans address none of them.
The shift
In one sentence
Google ranks pages. AI recommends entities. Two systems, two definitions of visibility, and most strategies address only the first.
For most of two decades that was fine, because the first system was practically the only one that mattered. Optimizing your site to rank well was a reliable proxy for people being able to find you, and the proxy held.
A second visibility system has emerged alongside it, and it runs on fundamentally different inputs. Most marketing teams sense something has moved. Very few have had a clean framework for naming what it is, which is what this is for.
The evidence
What changed, with the numbers behind it
This is no longer speculative. Each of these comes from a named study with a stated method, which matters, because a great deal of what circulates on this topic does not.
| Figure | What it measures | Source |
|---|---|---|
| -61% | Organic click-through rate on informational queries where an AI Overview appears, falling from 1.76% to 0.61%. | Seer Interactive, September 2025. 3,119 queries, 42 organizations, 25.1 million impressions. |
| -41% | Organic click-through rate on queries where no AI Overview appears at all. Clicks are falling everywhere, not only where AI is visible. | Seer Interactive, same study |
| 93% | of searches in Google's AI Mode end without any click to an external site. AI Mode replaces results rather than sitting above them. | Semrush, September 2025 |
| 35% | of US consumers use AI tools at product discovery, against 13.6% using traditional search at the same stage. | Similarweb, 2026 Generative AI Brand Visibility Index |
| 17 to 38% | Overlap between brands ranking in Google's top ten and brands cited in AI answers, down from roughly three quarters a year earlier. | Multiple studies, early 2026. Range varies by methodology. |
Now the row that argues for doing something rather than despairing. Seer also found that brands cited inside an AI Overview see roughly 35% higher organic click-through than uncited brands on the same results page.
So the picture is not simply that clicks are disappearing. Clicks are consolidating around whoever gets named. The gap between cited and uncited is widening on the same page, for the same query, at the same moment.
Why the lag
This is not a failure of your marketing team
It is that the shift outpaced the tools and reporting frameworks. Search Console does not tell you whether ChatGPT recommended you. Most search platforms have no field for AI visibility. The metrics teams report on, rankings and organic traffic and impressions, were built for the system that is now half the picture.
So a strategy built around those metrics is, by definition, optimizing for a shrinking share of how buyers actually find vendors. Not wrong. Incomplete. Those are different diagnoses with different responses, and the distinction is worth holding onto before anyone starts assigning blame internally.
The uncomfortable part: because the reporting has not caught up, the gap is invisible in exactly the documents where you would expect to find it. Your dashboards will look reasonable throughout.
The framework
The six things that determine AI visibility
If search visibility is measured by rankings, AI visibility is measured by something else entirely, and it breaks into six dimensions. Most strategies address none of them, not through neglect, but because the category did not exist until recently.
| Dimension | The question it answers | What failure looks like |
|---|---|---|
| Entity clarity | Do you exist as a clearly defined, consistently described entity in the knowledge layer these systems draw from? | Appearing on some phrasings and vanishing on others. The most common finding, and the cheapest to fix. |
| Content authority | Is your content specific enough to be worth quoting, or general enough that nothing would bother? | Pages full of sound advice that could have come from anyone in your category. |
| Structured data | Can an engine parse and extract cleanly, or does it have to interpret? | A beautiful site with no labels on any of it. Nothing breaks, which is why it survives for years. |
| Citation worthiness | Do you answer questions definitively, or do you hedge? | Answers so qualified that no clean sentence can be lifted from them. |
| Measured visibility | Are you actually being cited right now, across multiple platforms, for the queries that matter? | Nobody in the organization can answer this, and nobody has been asked to. |
| Technical foundation | Can AI crawlers reach and trust your site at all? | Everything above is irrelevant if this fails, which is why it gets checked first. |
Peer-reviewed work supports several of these directly. Researchers from Princeton and Georgia Tech, publishing at KDD 2024, tested nine content tactics across roughly ten thousand queries. Adding specific statistics produced the largest single lift at around 41 percent. Adding direct quotations came next at roughly 28 percent. Citing external sources helped substantially, particularly for lower-ranked pages.
Four of the nine tactics did nothing or actively hurt: keyword stuffing, simplifying for its own sake, padding content to length, and purely persuasive language. That list describes a great deal of what currently sits in most content guidelines.
Aggarwal et al., GEO: Generative Engine Optimization, KDD 2024. Reported lifts vary by metric and domain.
The sequence
This is not a content problem first
Which is the most important thing in this piece, because more content is always the first instinct and it is usually the wrong first move.
It is an infrastructure problem first. Before anything can be cited, the foundation has to exist. Entity signals consistent across your digital properties. Structured data implemented. Your site properly indexed in the places these engines actually retrieve from, which includes indexes you may not have thought about.
- 1
Infrastructure. Consistent entity signals, structured data, crawler access. Unglamorous, cheap, and load-bearing for everything after it.
- 2
Citation-worthy content. And even here it is not volume. It is specificity and depth in a focused area rather than broad coverage of everything.
- 3
Distribution and external presence. The part that takes longest and cannot be shortcut, because it depends on other people.
Reversing that order, which is what just produce more content effectively does, invests real effort into a system that cannot use it yet.
Keep going
Numbers with their methods attached.
What is actually changing in search and AI visibility, every figure sourced and dated, stated at its real size rather than its most persuasive one.
No spam. Unsubscribe anytime.
The implication
If your strategy has no line item for any of this
No visibility audit. No AI visibility score. No entity clarity assessment. That is not a criticism of the strategy. It is a description of when it was written relative to when this accelerated.
It does mean there is a gap between what the strategy measures and what increasingly determines whether buyers find you. And the overlap between the two systems keeps narrowing, from roughly three quarters a year ago to somewhere between seventeen and thirty-eight percent now, depending on whose methodology you use.
Worth one honest caveat, since this post is about measurement. That divergence is not uniform. Google's own AI answers still overlap substantially with Google's rankings, which makes sense given they run on the same systems. The sharp separation is on the platforms Google does not operate, and those are the ones growing fastest.
Common questions
The strategy gap, answered
How much do AI Overviews actually reduce clicks?
Seer Interactive measured a 61% fall in organic click-through on informational queries where an AI Overview appears, from 1.76% to 0.61%, across 3,119 queries and 25.1 million impressions to September 2025. Notably, queries without an AI Overview also saw a 41% decline, indicating clicks are falling generally rather than only where AI is visible.
What are the six dimensions of AI visibility?
Entity clarity, meaning whether you exist as a consistently described entity. Content authority, meaning whether your content is specific enough to quote. Structured data, meaning whether engines can parse it. Citation worthiness, meaning whether you answer definitively. Measured visibility, meaning whether you are actually being cited now. And technical foundation, meaning whether crawlers can reach and trust your site.
Should I produce more content to improve AI visibility?
Not first. This is an infrastructure problem before it is a content problem. Consistent entity signals, structured data, and crawler access have to exist before content can be cited. Only then does content become the lever, and even then the variable is specificity and depth rather than volume. Producing more content on a foundation that cannot support citation wastes the effort.
What content changes actually increase AI citation?
Research presented at KDD 2024 by Princeton and Georgia Tech researchers tested nine tactics across roughly ten thousand queries. Adding specific statistics produced the largest lift at around 41 percent, followed by direct quotations at roughly 28 percent, with citing external sources especially effective for lower-ranked pages. Keyword stuffing, content padding, and purely persuasive language produced no benefit or made results worse.
Is being cited in an AI answer worth anything if nobody clicks?
Yes, measurably. Seer Interactive found brands cited inside an AI Overview earn roughly 35 percent higher organic click-through than uncited brands on the same results page for the same query. Clicks are not simply disappearing. They are consolidating around whoever gets named, which widens the gap between cited and uncited businesses.
Start here
The starting point is a measurement, not an overhaul
The AI visibility audit tests your brand across ChatGPT, Perplexity, and Gemini, benchmarks you against your closest competitors, and identifies the specific structural gaps behind whatever it finds, ranked so you know what to do first. If your strategy was written before this accelerated, and for most companies it was, this is the fastest way to see exactly what changed.
Not ready yet? Subscribe to the newsletter and get the next breakdown in your inbox.