Nobody has fully cracked this yet. Here’s the honest state of play — and a practical recipe to start.
A friend of mine (owner of an e-commerce website) called me last month, slightly panicked.
His organic traffic had dropped around 20% since the start of the year. Not a penalty, not a technical issue. Just: the questions his blog used to answer were now being answered directly in ChatGPT, Gemini, and Google’s AI Overviews. Their visitors were getting the answer without ever clicking through.

The honest answer — which I gave him and which I’ll give you — is: we’re not entirely sure yet. But we’re not flying blind either. There’s enough evidence from early experiments, published research, and pattern observation to build a sensible working approach.
That approach has a name: GEO optimization. And it’s going to matter more than SEO for a lot of B2B teams in the next two years.
What GEO optimization actually is
GEO stands for Generative Engine Optimization.
SEO is optimising your content to rank in traditional search engines — Google, Bing, whatever. GEO optimization is about getting your content cited, referenced, or surfaced by AI systems when they answer a user’s question.
The difference matters. When you rank #1 on Google, people click your link, land on your page, you get traffic. When an AI mentions you in a response, the user might get their answer without ever visiting your site — but they heard your name, they saw your idea attributed to you, and some of them will come looking for you directly.
GEO optimization is less like a traffic tap and more like a reputation signal. Which, for B2B, is actually the channel that converts best anyway.
The scale of what’s at stake: according to Reuters citing internal OpenAI data, ChatGPT had over 900 million weekly active users as of February 2026 — more than double the 400 million reported in February 2025. (Source: DemandSage). That’s a lot of questions being answered without anyone clicking through to your site.
And it’s not just ChatGPT. Gemini is at 750 million monthly users. Perplexity, Claude, Copilot — the AI-answer layer of the web is enormous and growing.
Here’s the mechanism in plain terms. When someone asks ChatGPT, Gemini, or Claude a question, the AI draws on:
- Its training data (content absorbed before its knowledge cutoff)
- In some cases, real-time web search (Perplexity, ChatGPT with browsing, Gemini with Search integration)
- The quality and credibility signals it detects in that content
Your GEO optimization goal: be the source an AI considers trustworthy, specific, and citable when it constructs an answer in your area.
Why this is urgent
Before getting into the GEO optimization recipe, let’s be honest about the size of the problem.

Traditional organic search is under real pressure. According to Pew Research tracking 68,000 real searches from 900 U.S. adults (published July 2025), users clicked a result only 8% of the time when an AI summary appeared, versus 15% without one — a 46.7% relative click reduction. (Source: Demand Local)
And it’s not just AI Overviews. In Google’s newer AI Mode, 93% of searches end without a single click to an external website, per Semrush data from September 2025.
Gartner projects a 25% decline in traditional search volume by 2026 as generative AI search scales, with a 50%+ decline projected by 2028.
In this landscape, GEO optimization is not a nice-to-have. It’s the adaptation your content strategy needs.
Nobody has fully figured out GEO optimization
Let me be honest about something before we get into tactics.
GEO optimization is not like SEO circa 2018, where the rules were reasonably well understood and execution was mostly about effort and consistency. GEO in 2026 is more like SEO in 2003 — people are running experiments, publishing findings, observing patterns, and building working hypotheses. Some of those hypotheses will turn out to be wrong. The platforms themselves don’t publish their ranking criteria. The models update constantly.

The foundational research comes from a paper by Princeton, Georgia Tech, the Allen Institute for AI, and IIT Delhi — the paper that actually coined the term “Generative Engine Optimization.” Published at ACM SIGKDD 2024, the researchers tested 10,000 queries across multiple generative engines and evaluated nine different GEO optimization strategies.
The results: tactics like adding statistics, citing sources, and including quotable statements improved content visibility by up to 40% in AI-generated responses. Pages ranked around position 5 in traditional search saw a 115% visibility increase after GEO optimization. Pages already at position 1 saw minimal change.
That last finding matters: GEO optimization is a bigger opportunity for smaller, non-dominant sites than for established ones.
What this means practically: everything I’m about to give you is the best current working recipe, not guaranteed truth. Treat it as a starting point for your own experiments. Watch what gets cited. Test different formats. Stay curious.
That said — here’s the recipe.
The GEO optimization recipe
Think of GEO optimization in four layers: credibility signals, structure and clarity, topical authority, and off-site presence. Weak spots in any layer reduce your chances of being surfaced.
Layer 1: Credibility signals — give AI a reason to trust you
AI systems are assessing whether a source is trustworthy before surfacing it. The GEO optimization signals here are different from traditional SEO link authority, but the underlying question is the same: is this a credible source for this topic?
- Use real data, named with source URLs. Vague claims get ignored. Specific, sourced statistics get cited. “Many B2B buyers do research before contacting sales” is invisible. “According to Gartner, 75% of B2B buyers prefer a rep-free sales experience” is citable. Cite your sources inline, not just as hyperlinks — write the institution name and year in the text itself.
- Attribute everything you didn’t originate. AI systems trained on the open web have learned that properly attributed claims are more reliable than unattributed ones. This is one of the five GEO optimization tactics the Princeton paper specifically validated.
- Make your expertise explicit. Your About page, your author bio, your article introductions — these should make clear, in plain language, what you actually know and why. “I’ve run B2B demand generation in three countries across 14 years” is more credible to an AI parser than “I’m passionate about marketing.”
- Use your real name, consistently. Across your site, your LinkedIn, your guest articles. Named authorship is a credibility signal. Anonymous content is not.
Layer 2: Structure and clarity — write for something that reads, not skims
Content that is logically organized, clearly headed, and internally consistent is easier for AI to process and extract from. This is not an accident — it’s one of the core GEO optimization levers.
Answer the question directly at the top. Don’t make an AI wade through three paragraphs of context before getting to the substance. The Princeton GEO paper specifically validated “fluency optimization” — clear, direct writing — as a top-5 tactic with 30–40% visibility improvement. (Source: DerivateX)
Use explicit definitions. When you introduce a concept, define it. “GEO optimization (Generative Engine Optimization) is the practice of…” is more citable than assuming the reader knows. AI systems use definitional passages as anchor points when constructing explanations.
Write quotable sentences. Crisp, self-contained, meaningful out of context. “GEO optimization is less like a traffic tap and more like a reputation signal” is quotable. A long, hedged sentence with five qualifications is not. Every piece should have two or three sentences that could stand alone.
Use clear headers that mirror natural language questions. Not “Section 3: Tactical Considerations” — but “What actually works for GEO optimization right now?” Headers that match question patterns are more likely to align with the queries AI systems are trained on.
Layer 3: Topical authority — be the site that goes deepest on one thing
This is where the SEO and GEO optimization playbooks converge. AI systems favour sources that demonstrate deep, consistent expertise in a specific domain over sites that cover everything shallowly.
- Pick your territory and stay in it. Write ten articles on B2B lead generation from different angles before you write one article on something tangentially related. The pattern of your content matters as much as any individual piece.
- Cover a topic at multiple levels. Not just the overview article, but the specific tactical guide, the case study, the “what I actually found when I tested this” piece. If your site has the overview, the deep dive, and the practical example on a topic, you become a go-to source rather than a supplementary one.
- Update old content. Stale content signals the wrong thing. An article written in 2022 with no updates is less likely to surface in 2026 than one meaningfully revised recently. Add a “last updated” date that you actually honour.
- Fill the question gaps nobody else is filling. If you search a specific question in your domain and get five thin, generic articles — that’s your GEO optimization opening. Be the best available answer. AI systems surface the best available answer.
Layer 4: Off-site presence — get mentioned before you get cited
AI systems don’t just crawl your website. They’ve been trained on the entire open web — forums, news articles, academic papers, Reddit, LinkedIn posts, community discussions.
Your credibility in AI responses is partly built by what other parts of the internet say about you or quote you saying. This is the GEO optimization equivalent of link building — except it’s about mentions and citations, not just links.
- Get quoted in other people’s content. Guest articles, expert roundups, journalist quotes, podcast transcripts — every time your name and ideas appear on a credible external site, you’re building the off-site signal that AI training data picks up on.
- Be findable and quotable on LinkedIn. LinkedIn content is indexed and in some cases directly drawn on by AI systems for B2B topics. A detailed, well-argued post on your area of expertise can function as a citation source. Post consistently.
- Contribute substantively in communities. Reddit, Slack communities, industry forums — these are in the training data. A detailed, useful answer you wrote two years ago in the right subreddit might be contributing to your GEO optimization signal right now.
- Get on Wikipedia — or at least near it. Wikipedia is heavily weighted in AI training data. Being cited on Wikipedia, referenced in Wikipedia-adjacent sources, or quoted in articles that Wikipedia links to is meaningful for GEO optimization. Long game, but worth knowing.
What GEO optimization won’t do
Honest caveat, because this article would be dishonest without one.
GEO optimization is not a replacement for all other traffic sources. For high-intent, bottom-of-funnel queries — people who are ready to buy — traditional SEO still drives clicks better. GEO optimization is stronger for awareness and authority: getting your name and ideas into the heads of people who are earlier in their thinking.
One piece of good news for smaller sites: The HOTH’s data shows that visitors who arrive through AI referrals convert at 9x the rate of standard organic traffic. (Source: The HOTH) The volume is lower, but the intent is high.
GEO optimization also doesn’t work on a short timeline. The content you improve today may not affect your citation rate for six months. This is a compounding game, not a quick win.
And nobody, including me, has a complete and verified understanding of how these systems weight sources. We have working hypotheses built from observation and early research. Treat the recipe above as a high-probability bet, not a guarantee. Run your own experiments. Keep notes. Share what you find.
The tactics — credibility signals, clear structure, topical depth, off-site presence — are not revolutionary. They’re a more rigorous version of what good content has always required. What GEO optimization actually demands is a shift in why you create content.
If you’ve been creating content to capture traffic, GEO optimization forces you to think about creating content to build reputation. Those are not the same goal. Traffic is a number on a dashboard. Reputation is the reason someone types your name into an AI when they’re trying to solve a problem.
The B2B companies that will win at GEO optimization are the same ones winning at organic growth generally: those who go deep on a specific topic, build real expertise, share it consistently, and earn trust over time.
The AI raises the stakes. And it rewards exactly the things shortcuts and average content never could.

Comments