By Paul Okhrem · paul-okhrem.com
We spent three months running a structured research project to answer a question that almost every B2B marketing leader I talk to is asking but nobody has clean data on: how visible are B2B brands when buyers use AI to do research?
The results are available for free. This post summarizes the key findings and the methodology. The full report is at paul-okhrem.com.
Why this research exists
Generative Engine Optimization — GEO, or sometimes AEO for Answer Engine Optimization — is the practice of improving how a brand appears in AI-generated responses, not just in traditional search rankings.
The concept has been discussed for two years, but reliable benchmarks have been hard to come by. Most of the content in this space is either:
- Vendor-produced, with obvious selection bias toward cases where optimization worked
- Too generic to be actionable for specific verticals
- Based on consumer contexts rather than B2B buying scenarios
We focused specifically on B2B: industrial, technology, professional services, and manufacturing sectors. We tested 200+ companies across four verticals using a standardized prompt protocol across three major AI platforms.
How we measured visibility
GEO visibility is harder to measure than search ranking for several reasons. There's no equivalent of a rank position. AI responses are generated dynamically. The same query can produce different responses across sessions. And the factors that drive inclusion in AI responses are not fully transparent.
Our methodology:
Query design: We built a library of 80 buyer-intent queries across four stages — problem recognition, solution exploration, vendor comparison, and decision validation. Each query was designed to mirror real buyer language, not marketing language. "What's the best ERP for a mid-size manufacturer?" rather than "top enterprise resource planning solutions."
Platforms tested: ChatGPT (GPT-4o), Claude (Sonnet), and Perplexity. We tested each query 3 times per platform over a 6-week period to account for variation.
Visibility scoring: We scored each company's appearance in responses across four dimensions: mentioned at all, mentioned by name in a relevant context, mentioned with specific detail (capability description, differentiator), and mentioned with a source citation. Companies could score 0–4 per query.
Normalization: We normalized scores by query type and platform to allow cross-company comparison.
Key findings
1. AI visibility and traditional search ranking are correlated but not equivalent.
Companies with strong SEO tended to have higher AI visibility — but the correlation was approximately 0.6, not 0.9. A meaningful portion of companies ranked well in traditional search but poorly in AI responses, and vice versa. This suggests that AI visibility requires distinct optimization work, not just better SEO.
2. Content depth drives AI inclusion more than content volume.
Among the companies in our sample, those with high AI visibility consistently had deeper content on specific topics — detailed technical documentation, structured comparison pages, specific use-case content — rather than more pages overall. The companies with the most content weren't the most visible in AI responses.
3. Third-party mentions matter significantly.
Brands that appeared in AI responses with specific detail were almost always mentioned in credible third-party content — analyst reports, industry publications, detailed review platforms. Self-published content alone rarely drove the kind of AI visibility that includes specific detail or differentiating claims.
4. Problem-stage queries favor incumbents; solution-stage queries are more open.
When buyers are at the "I have this problem" stage, AI responses tend to mention established category leaders. At the solution exploration and vendor comparison stages, there's more room for challengers to appear — particularly if they have strong, specific content on the comparison terms buyers use.
5. Industrial and manufacturing verticals have the most GEO opportunity.
Compared to tech and professional services, industrial and manufacturing companies had the lowest AI visibility scores despite significant buying activity happening in these sectors. The content gap is real, and the competition for AI visibility is lower. This is where the opportunity is largest for companies willing to invest now.
What the research doesn't cover
A few important limitations:
This is a point-in-time study. AI models are updated continuously, and what drives visibility today may shift. The benchmark is useful as a baseline; it's not a static truth.
We measured visibility, not conversion. Whether AI visibility translates to pipeline is a separate question we didn't study — and it should be studied, because visibility that doesn't drive buyer behavior is ultimately vanity.
The query library reflects our judgment about what buyers actually search. It's based on interviews with 40 B2B buyers across the four verticals, but it's not exhaustive.
How to use the benchmarks
The full report includes:
- Per-company visibility scores for all 200+ companies in the sample (anonymized by company except for those who opted into attribution)
- Vertical-specific benchmark distributions — so you can see where your score would rank in your actual competitive set
- Query-level data — which types of queries your company appeared in, and which you didn't
- A self-assessment checklist for GEO readiness based on the factors that correlated with high visibility
If you want to understand where your brand stands before investing in GEO optimization, the benchmark data is the place to start.
The full research is available at paul-okhrem.com.
Paul Okhrem publishes practical research on AI visibility and B2B marketing. More at paul-okhrem.com
Top comments (0)