Expert Take: Adapting to hyperpersonalisation is the biggest strategic challenge for B2B marketers right now — though most of them haven’t realised it yet, I think.
What Is GEO — and Why Should B2B Marketers Care?
GEO stands for Generative Engine Optimization. The term comes from a 2024 study by Princeton University and Georgia Tech, and describes the process of optimizing content so that AI search engines — ChatGPT, Perplexity, Google AI Overviews, Claude — select and cite you as a source.
That sounds abstract, but the numbers make it concrete. Gartner predicted in 2024 that traditional search engine volume would drop 25% by 2026. That seemed like a bold prediction at the time. Now, in early 2026, I see it in the data: AI-referred website visits increased 527% in five months — from 17,076 to 107,100 sessions — according to Previsible’s 2025 AI Traffic Report.
For B2B marketers, there’s an additional reason to take this seriously. According to a 6sense study covering 680 million citations, 73% of B2B buyers use AI tools during purchase research. In the tech sector it’s even more extreme: 80% of tech buyers use AI tools at least as much as traditional search engines, and 56% discover vendors through chatbots.
The number that changed my thinking most: AI search traffic converts 5.1 times better than organic Google — 14.2% versus 2.8%. That shifted my sense of urgency around GEO.
How Do AI Search Engines Select Their Sources?
This is where most online advice falls short. Because each AI platform cites in a fundamentally different way.
Perplexity Leans Heavily on Reddit
Perplexity draws 46.7% of its top citations from Reddit, according to research by Discovered Labs. That’s nearly half. If you want visibility in Perplexity and you have no Reddit strategy, you’re missing half your potential reach.
Why Reddit? Because it contains authentic discussions with personal experiences — exactly the type of content AI systems find valuable. Reddit contributes 40.1% of all LLM training data, more than Google and Wikipedia. Google signed a $60 million licensing agreement for that data in 2024.
ChatGPT Trusts Wikipedia and Competitor Sites
ChatGPT cites Wikipedia in 47.9% of cases, according to The Digital Bloom’s 2025 AI Citation Report. More striking: ChatGPT cites competitor websites 11.1 percentage points more often than Google does. That means your competitors can become more visible via AI search than through traditional search results.
Claude Prefers Technical Precision
Claude — Anthropic’s AI model — cites content that is structured with bullets and clear headings 30% more often, according to Discovered Labs. That’s relevant because Claude is increasingly used for business research.
The Overlap Is Surprisingly Small
Perhaps the most telling data point: only 11% of domains are cited by both ChatGPT and Perplexity. Each platform largely has its own source set. That makes “generic GEO advice” a contradiction in terms — you need to understand what works per platform.
What Factors Determine Whether AI Cites You?
The Princeton/Georgia Tech study identified several factors that increase AI visibility. I combined those with more recent research to build a more complete picture.
Statistics and Citations: The Biggest Levers
Adding verifiable statistics to your content increases AI visibility by 22–40%, according to the Princeton/Georgia Tech study. Expert quotes deliver 41% more visibility. Inline citations — source references in the text itself — add another 30%.
Those aren’t subtle improvements. It means an article with supported claims and source references has nearly 50% more chance of being cited than the same article without those elements.
Structure Matters More Than Length
More than 50% of AI-cited pages contain fewer than 1,000 words, according to analysis by Bradley Lee Bartlett. Structure predicts citations better than length. Specifically: pages with 120 to 180 words between headings get 70% more ChatGPT citations than pages with sections under 50 words.
I found that striking. It means the trend toward longer, more comprehensive content we know from SEO doesn’t necessarily translate to GEO. What matters is whether an AI system can quickly distill a relevant answer from your content.
Authority Works Differently Than You Think
In traditional SEO, backlinks are a key ranking signal. In GEO, that’s much less the case. Semrush found that brand search volume — how often people search directly for your brand — correlates 0.334 with AI citations, while organic keywords score 0.41 and backlinks lag at 0.37.
But the most striking finding about authority comes from Authority Tech: content on your own site has a citation ratio of 8%. That same content, distributed via news media, reaches 34% — a 325% improvement. Earned media — publications about you on other sites — is far more powerful in GEO than in traditional SEO.
Freshness Matters
Pages updated within 60 days are 1.9 times more likely to appear in AI answers, according to The Digital Bloom. That means content maintenance in GEO isn’t a nice-to-have — it’s a core activity.
What If AI Gets Something Wrong About You?
Most GEO advice assumes you want to be cited. But what do you do if an AI tool spreads incorrect information about your brand? I couldn’t find a good answer to that question anywhere.
LLMs like ChatGPT and Claude are trained on a snapshot of the web. There’s no mechanism to send a correction directly to a trained model — no “report an error” button that actually works. What you can influence are the sources those models rely on. ChatGPT cites Wikipedia in 47.9% of cases — if the information about you there is correct, you increase the chance of an accurate representation. Wikidata, which Wikipedia relies on for structured data, is a second anchor point that too few B2B marketers know about.
For systems with real-time retrieval — Perplexity and Google AI Overviews — the situation is more favorable: they pull information from current web content. A quick correction on your own site and on authoritative external sources can have an effect there.
In practice: monitor what AI tools say about you. Tools like Otterly.AI now offer brand monitoring specifically for AI answers. Publish corrections on authoritative platforms, not just your own site. And build a consistent information foundation: a correct Wikidata profile, verifiable information across multiple sources, a clear author page with credentials. That’s your best defense against incorrect citations — not a guarantee, but a solid base.
What Does the Zero-Click Trend Mean for B2B?
69% of Google searches now result in zero clicks, up from 56% earlier. For searches that trigger an AI Overview it’s even more extreme: 83% zero-click.
At the same time, the composition of those AI Overviews is shifting. In early 2025, 76% of citations in AI Overviews came from top-10 search results. By early 2026, that dropped to 38%. Google’s AI is increasingly sourcing from outside the first page.
For B2B marketers, this means two things. First: if your strategy revolves around “getting to page one,” you’re chasing the wrong metric. AI systems select sources on different criteria than search result position. Second: zero-click isn’t necessarily bad. If your brand is consistently mentioned as a source in AI answers, you’re building brand recognition — even without a click.
How Does GEO Differ from SEO — and Where Do They Overlap?
This is a question I asked myself when I started my project. The short answer: they overlap more than most GEO articles suggest, but the emphasis is different.
Where They Overlap
Google SEO rankings correlate 0.65 with LLM visibility, according to Otterly.AI research. That’s a strong correlation. A site that ranks well in Google has a head start in AI visibility — but it’s not a guarantee.
Technical SEO basics — fast load times, mobile-friendly, structured data — also help for GEO. AI crawlers struggle with JavaScript-heavy content just like Google does. Schema markup as static JSON-LD is relevant for both.
Where They Fundamentally Differ
The key differences are in three areas:
Citation versus ranking. In SEO you want to rank high in a list of ten results. In GEO you want to be cited as part of an answer. That requires content that gives a concrete answer to a specific question, not content that broadly optimizes for a search term.
Authority versus popularity. Backlinks — the currency of SEO — correlate only 0.10 with LLM visibility, according to Otterly.AI. What does matter: brand authority, earned media, and the quality of your source references.
Multimodal versus text. Multimodal content — content with infographics, diagrams, data visualizations — has a 156% higher chance of being selected by AI. In SEO, visual content is a nice-to-have; in GEO it’s becoming a strategic factor.
What Does a Practical GEO Plan Look Like for B2B?
I can’t give you a one-size-fits-all step plan — every situation is different. But based on what I’ve learned in my own project, these are the areas I would address.
Start With Your Existing Content
You don’t have to start from zero. Most B2B sites already have content that becomes more GEO-friendly with relatively small changes:
Add verifiable statistics with source references. At least one concrete, verifiable data point per paragraph. Structure your headings as questions that a B2B buyer would ask an AI search engine. Add a TL;DR at the top — AI systems often use that summary as the basis for their answer. Check that your schema markup is correct and delivered as static JSON-LD. That sounds technical, and it is. Like SEO, GEO has a technical component. If you can’t handle that yourself, make sure you find someone who can.
Invest in Authors, Not Just Content
AI systems value recognizable authors. An article from “the marketing team” carries less weight than an article from a person with demonstrable expertise. Make sure your author pages have relevant credentials, that articles list a clear author, and that author is findable outside your own site — on LinkedIn, in trade publications, at conferences.
Build a Platform-Specific Strategy
Because each AI platform cites differently, your strategy needs to differ per platform. For Perplexity: invest in Reddit presence with authentic, valuable contributions — not link-dropping. For ChatGPT: make sure your Wikipedia mentions are accurate and that your content appears on high-authority domains. For Google AI Overviews: focus on structured data and direct answers to frequently asked questions.
Earned Media Is Your Biggest Lever
With a citation ratio of 34% via news media versus 8% on your own site, earned media is the most powerful GEO strategy most B2B marketers aren’t yet using. That means: guest articles at relevant trade publications, quotes in industry research, contributions to podcasts and webinars. Not for the backlinks — for the AI visibility.
Measure What You Do
Measuring GEO is harder than measuring SEO, but not impossible. Tools like Otterly.AI, HubSpot’s AI Search Grader, and Semrush’s Copilot now offer dashboards for AI visibility. At minimum, measure: which AI answers you appear in, which pages get cited, and how your AI-referred traffic develops.
Decide What You Share with AI Crawlers
GEO assumes you want to be cited — but that doesn’t apply to all content. Confidential price lists, exclusive whitepapers, or customer case studies you’d rather not see in AI training data can be blocked via your robots.txt. The major AI platforms have their own crawlers: OpenAI uses GPTBot and OAI-SearchBot, Common Crawl (training data for nearly all major LLMs) has CCBot, Perplexity uses PerplexityBot. Correct robots.txt instructions are respected by these platforms.
The nuance most marketers miss: you can block the crawler for training while allowing the crawler for search retrieval. OpenAI deliberately split this into GPTBot (training) and OAI-SearchBot (ChatGPT Browse). Make sure you decide which content you leave open and which content you shield.
How Do You Measure AI Conversions When Traffic Is Invisible?
If AI search traffic converts so much better, you want to know how many of your leads come through that route. But a large portion of that traffic is what marketers call “dark traffic”: sessions that appear in GA4 as “direct / none” because the AI tool doesn’t pass referrer information.
Partially this is measurable. ChatGPT Browse and Perplexity clicks increasingly appear in GA4 as recognizable referrers — chatgpt.com and perplexity.ai respectively. Claude refers via claude.ai. A custom report in GA4 filtering on these domains gives you a lower bound of your AI-referred traffic. Not the complete picture, but a start.
Tools like Otterly.AI measure how often your brand and pages are cited in AI answers — that gives an indication of visibility, even when the click-through isn’t traceable. And then there’s the most low-tech solution: a “how did you find us?” question in your contact form, with AI chatbot as an explicit option. It sounds inelegant for people used to dashboards, but in my experience it delivers reliable qualitative data that no analytics package gives you. The most reliable source often remains the customer themselves.
GEO in Non-English Markets: Why Language Matters More Than You Think
Most of the GEO research I cite in this article is based on English-language content and queries. That’s relevant, because LLMs are predominantly trained on English data — training density for other languages is considerably lower. In practice, I see this play out: for the same search query in Dutch and English, an AI tool sometimes gives noticeably different answers, with different sources and different emphasis.
What that means for B2B marketers in non-English markets: authoritative sources in your language carry relatively more weight in language-specific queries, simply because there are fewer competing sources. A mention in a leading national trade publication has more weight in a local-language AI query than the same mention in an English-language niche medium — because there are fewer alternatives.
That makes early investment in local-language authority platforms attractive. You fill a gap your English-speaking competitors can’t fill. I haven’t been able to find good comparable research that directly compares GEO performance across languages — that’s honestly a blind spot in the field. But it follows logically from what we know about how LLMs train, and it’s a hypothesis I’m testing in my own experiments.
What’s Still Unclear — and Why That’s Interesting
The GEO discipline is young. The Princeton/Georgia Tech study is only from 2024. Most statistics I cite in this article are from 2025 — they’re current, but the trends are moving fast. The field is in full development.
And there’s something on the horizon that most GEO articles don’t mention: AI agents. In 2026, most organizations are still focused on getting found in AI answers. But the next step — AI agents that autonomously book demos, compare quotes, or fill out contact forms on behalf of a buyer — is probably going to arrive faster than expected. What that means for GEO: it won’t just be about being cited in an answer, but about technical accessibility for automated interaction. Forms that are machine-readable, APIs that agents can call, contact flows that work without captcha barriers. I don’t yet know exactly how this will play out, but it’s a development that could rewrite the rules of GEO again.
I also don’t know how AI platforms will adjust their algorithms as more marketers optimize for AI citations. We don’t know how the legal discussions will pan out — Reddit filed a lawsuit against Anthropic in June 2025 over unauthorized use of training data, and The New York Times did the same against the major AI companies.
What I do know: the direction is clear. 73% of B2B buyers already use AI tools for purchase research. 65% expect to increase that usage. This is not a trend that will blow over.
Frequently Asked Questions About GEO for B2B
Is GEO a replacement for SEO?
No. Google SEO rankings correlate 0.65 with AI visibility — a well-optimized site has a head start. GEO is a complement that responds to how AI search engines select sources. The two reinforce each other.
How quickly does GEO deliver results?
Faster than traditional SEO in some respects. Because AI systems evaluate content on quality and relevance rather than link profiles and domain age, you can become visible more quickly with the right content. But building brand authority — an important factor in GEO — takes time.
Is GEO only relevant for tech companies?
No, but adoption is furthest along there. 80% of tech buyers use AI tools for vendor research, versus lower percentages in other sectors. The expectation is that other B2B sectors will follow — the question isn’t if, but when.
Which AI search engine is most important for B2B?
That depends on your target audience. ChatGPT has 80.5% market share, but Perplexity is growing fast and is widely used for business research. Google AI Overviews reach the broadest audience. My advice: start with the platforms your target audience uses and expand from there. You may already be able to see this in your dashboards.
How much does a GEO strategy cost?
The basic elements — enriching content with statistics, improving structure, adding schema markup — can be done with existing resources. A complete GEO strategy including earned media, platform presence, and platform-specific optimization requires more investment. GEO as a service is still rare at most agencies, but it’s only a matter of time before that changes.
What do I do if an AI spreads incorrect information about my brand?
There’s no direct way to correct a trained LLM. What you can do: improve the information in the sources AI systems rely on, such as Wikipedia and Wikidata. For systems with real-time retrieval (Perplexity, Google AI Overviews), publishing corrections quickly on your own site and on authoritative external platforms can help. Monitor actively what AI tools say about you — tools like Otterly.AI now offer specific functionality for this.
Does GEO work differently in non-English markets?
The dynamics differ. LLMs are predominantly trained on English data, which means training density for other languages is lower. Authoritative sources in your language therefore carry relatively more weight in language-specific queries, simply because there are fewer competing sources. Early investment in those platforms can deliver a relatively large advantage.
Sources
All statistics in this article with their sources:
| Statistic | Source |
|---|---|
| 527% increase in AI traffic (Jan–May 2025) | Previsible 2025 AI Traffic Report, via Search Engine Land |
| 25% drop in traditional search volume by 2026 | Gartner, February 2024 |
| 73% of B2B buyers use AI tools | 6sense Buyer Experience Report 2025 |
| 80% of tech buyers use AI as much as search engines | Forrester/Digital Commerce 360, 2025 |
| AI traffic converts 5.1x better (14.2% vs 2.8%) | 6sense Buyer Experience Report 2025 |
| 46.7% of Perplexity citations from Reddit | Discovered Labs, 2025 |
| 47.9% of ChatGPT citations from Wikipedia | The Digital Bloom 2025 AI Citation Report |
| 11% domain overlap between ChatGPT and Perplexity | Superprompt.com, 2025 |
| 40.1% of LLM training data from Reddit | TechnoSports, 2025 |
| Statistics increase AI visibility 22–40% | Princeton/Georgia Tech GEO study |
| Expert quotes increase visibility 41% | Passionfruit, 2025 |
| 50%+ of cited pages < 1,000 words | Bradley Lee Bartlett, 2025 |
| 120–180 words between headings = 70% more citations | The Digital Bloom, 2025 |
| Brand search volume correlation 0.334 | Semrush, 2025 |
| Earned media: 34% vs 8% citation ratio | Authority Tech, 2025 |
| Pages updated < 60 days: 1.9x more AI citations | The Digital Bloom, 2025 |
| 69% zero-click search results | Similarweb/Stan Ventures, 2025 |
| 83% zero-click on AI Overviews | Similarweb/Stan Ventures, 2025 |
| AI Overview citations from top-10: 76% → 38% | ALM Corp, 2026 |
| SEO rankings correlate 0.65 with LLM visibility | Otterly.AI, 2025 |
| Backlinks correlate 0.10 with LLM visibility | Otterly.AI, 2025 |
| Multimodal content: +156% selection chance | Otterly.AI, 2025 |
| 80.5% ChatGPT market share | First Page Sage, January 2026 |
| 65% of B2B buyers expect to use more AI | Marketing Charts, 2025 |
| GPTBot (training) vs OAI-SearchBot (retrieval) | OpenAI, GPTBot documentation |
This article was written by Hans Schepers, founder of Hands on GEO. More about my background on my author page.
Read also: What is GEO? · What is Entity Optimization?