Search interest in GEO, AEO, and related disciplines has grown sharply over the past year — but search volume is a proxy for curiosity, not a measure of results. What follows is what the data actually supports, what it doesn’t, and what to do with it.
Table of contents
Key takeaways
- Search demand for generative engine optimization reached 7,200 searches per month in the US, up 84% year over year, according to Ryan Law, Director of Content Marketing at Ahrefs, writing in September 2026.
- Demand for “ai visibility tools” grew 266% year over year — the steepest rate in the cluster — but the absolute volume is not reported, which limits interpretation.
- An Ahrefs analysis of one million pages across 100,000 search results found AI-generated content appearing at position one; the authors characterize the quality gradient as a quality effect, not a penalty for AI authorship.
- Every keyword figure in this article comes from Ahrefs Keywords Explorer, a single vendor source with known structural limits.
- No site-level outcome data — clicks, referral sessions, revenue — is attached to any of the trends described. These are demand signals, not performance benchmarks.
What this is about
Ryan Law, Director of Content Marketing at Ahrefs, put out a trend review in September 2026 covering eight shifts he observed across AI-influenced marketing. The figures discussed here come from Ahrefs Keywords Explorer US search volume data across roughly a dozen keyword clusters, spanning job titles, content tactics, and emerging optimization disciplines. No sample size for the keyword dataset is stated.
The design
This is an observational review, not a controlled experiment. Law pulled keyword demand data from Ahrefs Keywords Explorer — a tool his employer builds — and used year-over-year volume changes as a proxy for practitioner interest in each trend. For most YoY figures the comparison period appears to be September 2025 to September 2026, though exact date windows aren’t stated for every term.
No control group, no randomization, and no outcome variable beyond search volume itself. Calling this a study would overstate it; it’s a structured reading of demand signals from one keyword database. The Ahrefs content-ranking analysis (one million pages, 100,000 search results) is a separate data pull with its own methodology, briefly described but not fully documented in the source.
The numbers
The table below reproduces the figures Law reports, grouped by cluster. Where volume or growth is not stated in the source, the cell is left blank rather than estimated.
| Keyword | Monthly US volume | YoY change | Notes |
|---|---|---|---|
| generative engine optimization | 7,200 | +84% | Core GEO discipline term |
| ai visibility | 4,400 | +145% | Broad category term |
| ai visibility tracker | 2,400 | +156% | Tool-category term |
| ai visibility tools | not stated | +266% | Highest growth rate in cluster |
| llm visibility | 1,200 | +126% | — |
| agentic ai | 111,000 | +37% | Largest absolute volume in source |
| ai agents | 46,000 | +17% | — |
| agentic workflow | 2,300 | +89% | — |
| agentic commerce | 6,400 | not stated | — |
| ai image generator | 361,000 | +16% | Largest content-tool term |
| ai video generator | 223,000 | +21% | — |
| ai writing | 17,000 | +14% | — |
| best ai writing tools | 2,100 | +35% | — |
| ai ugc | 3,400 | +36% | — |
| ai influencers | 1,800 | +12% | — |
| ai avatars | 900 | +13% | — |
| marketing ai engineer | 0 | first seen 2026 | Emerging job title |
| head of ai marketing | not stated | first seen 2025 | Emerging job title |
| claude code | 508,000 | not stated | Build-vs-buy proxy |
| vibe coding | 86,000 | from near zero in a year | — |
No confidence intervals are provided for any figure. No statistical significance tests are reported. These are point estimates from a proprietary keyword database.
What could still be wrong
Several issues limit how far these numbers travel.
Single-source vendor data. Every keyword figure originates from Ahrefs Keywords Explorer, a product Ahrefs sells. Law is Ahrefs’ Director of Content Marketing. That is a direct conflict of interest, and readers should weight the figures accordingly. The methodology for volume estimation — panel data, clickstream, modeled extrapolation — is not disclosed in the source. Different keyword tools will often produce meaningfully different volume estimates for the same term.
Search volume measures curiosity, not adoption. A term growing 84% year over year tells you more people are searching for it. It does not tell you how many practitioners have implemented GEO, whether those implementations drove measurable referral traffic, or whether the people searching converted into buyers of anything. The demand signal and the business outcome are separate questions.
Missing denominators on the growth rates. A 266% year-over-year increase means very little without the base volume. If “ai visibility tools” had 100 monthly searches a year ago, 366 searches today is noise. Law does not report the prior-year absolute volume for the highest-growth terms, which makes those percentage changes uninterpretable as magnitude claims.
Seasonality is unaddressed. A September 2026 publication date means the YoY comparisons land on September 2025 as the base month. If September 2025 was an unusually low-demand month for any of these terms — perhaps because a competing concept had not yet been named — the growth rates are inflated relative to a 12-month rolling average.
The content-ranking analysis has an undisclosed methodology. The claim that AI-generated content appears at position one comes from an Ahrefs analysis of one million pages across 100,000 search results. How AI-generated content was identified — classifier, manual labeling, disclosure tags — isn’t explained in the source. Misclassification in either direction would change the finding materially.
Segment mix in keyword clusters. Broad terms like “ai visibility” aggregate very different searcher intents — brand managers, SEO practitioners, journalists, students. A 145% growth rate on an aggregated term may reflect a single new audience segment entering the query space, not uniform growth across all prior segments.
What to do on Monday
If you want to run the same demand-signal check on your own priority topics, the method is straightforward regardless of which keyword tool you use:
1. Pull monthly volume for your target terms for the current month and the same month one year prior. Use the same tool for both periods — switching tools mid-comparison introduces methodology variance, not real change. 2. Record the absolute volumes alongside the percentage change. A term moving from 50 to 150 searches per month is a 200% increase and probably irrelevant to your planning. Moving from 3,000 to 4,500 is a 50% increase and probably worth your attention. 3. Look at the full 12-month trend line rather than just the two endpoints — a single-month comparison can land on a seasonal trough or peak and misrepresent the underlying direction. 4. Separate the demand signal from the outcome question. Rising search volume for “generative engine optimization” tells you the discipline is attracting attention. To know whether GEO efforts produce referral traffic from AI systems for your site specifically, you need referral data from your own analytics. Treat the two as separate measurements.
For the AI-content-ranking question specifically: testing whether AI-assisted pages on your own site perform differently from fully human-written ones requires a documented production log that records which pages used AI generation and to what degree, a clean date of publication, and enough pages in each group to reach statistical power before you draw any conclusions. “We started using AI in March and rankings went up” is not an experiment — it’s an uncontrolled before/after comparison, and it doesn’t answer the question.
FAQ
Is search volume for a discipline term a reliable proxy for practitioner adoption?
No, and the gap matters. Search volume captures the population asking questions about a topic. That population includes people who’ve never tried the practice, students writing papers, journalists covering trends, and practitioners at every skill level. High volume growth can precede widespread adoption, coincide with it, or reflect media coverage of a concept that few people are actually implementing. To measure adoption, you need survey data with a defined sample, job posting data with a consistent methodology, or direct measurement of outputs — none of which are present in the source reviewed here.
How should I interpret a year-over-year growth rate when the base volume is missing?
With significant caution. A percentage change without its base is a ratio with one number hidden. Ask yourself what the absolute volume would need to be to matter for your planning. If “ai visibility tools” at +266% represents fewer than a few hundred monthly searches, it’s below the threshold where most businesses would prioritize content or channel investment. Get the prior-year absolute volume before treating any high-growth percentage as a strategic signal. When a source omits the base, that omission should be noted explicitly rather than papered over with the growth rate alone.
What would a more rigorous version of this analysis look like?
A stronger version would report absolute volumes for both periods rather than percentages alone, specify the exact date windows used for each YoY comparison, use at least one independent keyword database as a cross-check, separate branded from non-branded query intent within each cluster, and attach at least one outcome metric — referral sessions, conversion events, or indexed-page counts — to connect the demand signal to something downstream. For the content-ranking analysis specifically, a more rigorous version would document the AI-detection methodology, report inter-rater reliability if human labeling was involved, and publish the distribution of AI-use levels rather than only the top and bottom of the gradient.
For what the referral side looks like, see LLM referral traffic: what 13 months of conversion data actually show.
