In this episode, Demand Genius founder Tom Rudnai, reveals to Babel's Ash Lockett why B2B brands must stop treating AI like traditional SEO. To succeed at Generative Engine Optimisation (GEO), marketers must provide genuine information gain through unique perspectives, while clearing outdated 'content debt' to ensure AI accurately understands and recommends their B2B tech brand.
LINKS
Tom Rudnai (LinkedIn) https://www.linkedin.com/in/tom-rudnai-0539b6151/
Ash Lockett (LinkedIn) https://www.linkedin.com/in/ashlockett/
Babel Website: https://babelpr.com/
Demand Genius Website: https://demand-genius.com/
Ash Lockett: Hi, and welcome to Babel's Back to Brave podcast, the podcast that supports B2B tech communications and marketing professionals to make brave moves. I'm your host, Ash Lockett, Senior Director of B2B Marketing at Babel, and today I'm joined by a guest that's helping transition B2B marketing from outdated SEO playbooks to true Generative Engine Optimization (GEO): Tom Rudnai, founder of Demand Genius. Welcome, Tom, and thanks for joining me.
Tom Rudnai: Thank you. Lovely to be here.
Ash Lockett: So, for the listeners that might not know you—obviously I know you quite well, as I've been talking a lot to you about GEO and what it means to the B2B market—could you give us a quick rundown of who you are and what Demand Genius is?
Tom Rudnai: Yes. I mean, we've known each other about ten years now, so you've got the full background. I'm Tom Rudnai, founder of Demand Genius. My background was actually in the publishing space. I spent the majority of my career in AI and tech companies across sales, marketing, and ops roles, often working with publishers. I helped them understand how to get the most value out of their content in terms of balancing subscription and ad revenue.
That experience took us down the route to what we do now, which is helping B2B teams get the most out of their content distribution. And obviously, as we did that, this thing called AI took off and changed everything. That's what dragged us into this very fun, very interesting, and highly changeable world of GEO—trying to influence when and how AI talks about your brand.
That’s what we do at Demand Genius, specifically with B2B brands. We found GEO—or AEO (Answer Engine Optimization), whatever your favorite acronym is—to be a very different challenge in the B2B space. It’s a lot more meaty. It moves away from just optimizing visibility at the point of transaction, and towards asking: how are you positioning yourself, and how does AI reflect that back to buyers?
Ash Lockett: It definitely gets more complex within the B2B space, particularly when you start thinking of buying committees and their own LLM (Large Language Model) context in which they're searching.
I want to have a look at some of the industry misconceptions, because there are a dime a dozen sellers out there trying to sell you an audit or a visibility tool. One of the big common misconceptions we're seeing is that a lot of companies are starting to treat AI exactly like a traditional search engine. It’s all about: How do I get cited? How do I replace my blue-link clicks with bottom-of-the-funnel citations?
So my first question to you is: Is the marketing world treating GEO exactly like SEO? And why is it dangerous for B2B tech brands to do so?
Tom Rudnai: They are treating it like SEO, and the simplest answer is because they want it to be SEO. We want AI to just be this next iteration of search because it means we can solve it with our existing playbooks and structures without doing the hard work of unpicking it.
Partly for that reason, this was a problem first picked up by the SEO industry. Within months, citation tracking emerged as a way of measuring AI visibility, and honestly, there's been almost zero innovation in it since. That was the playbook SEO folks generated because they were coming at it from their own standpoint.
But here is my more meta answer: we're trying to solve the problem of AI visibility using corporate structures that don't allow for it. AI is a reflection of your entire digital footprint. Yet, we operate in a world where our teams are built around channel specialisms. You end up with PR people, SEO people, and content people who all swear blindly that they hold the single key to getting you cited in ChatGPT. They're all a little bit right, and a little bit wrong.
Ash Lockett: I think the SEO industry led the charge because they saw their Google traffic drop overnight once AI modes were implemented, so they tried to find quick fixes to replace that pipeline.
But I know you recently conducted some research on this. You found that AI answers from its training memory about 84% of the time during the awareness and consideration phases, and only retrieves external information via search about 16% of the time right at conversion. What is happening in that 84% of the journey that marketers are missing?
Tom Rudnai: Let’s take a step back and give the context of that study. We were analyzing how AI responds at the awareness, consideration, and conversion stages of the funnel.
Anecdotally, you see the language change depending on the stage. At the start of a user's prompt journey, the LLM is very exploratory. It then becomes comparative, and finally, very directive. It goes from exploration mode into helping you make a decision. We call it "intent matching."
Citation and live retrieval is basically 0% in awareness, 0% in consideration, and around 48% at the conversion stage. So, when you do SEO-led GEO, you are exclusively optimizing for that 48% of queries right at the bottom of the funnel where AI invokes retrieval.
Ash Lockett: That’s fascinating. A B2B buyer at the top of the funnel is trying to understand their challenges, so the LLM is educating them. As they build their thesis, it prompts them down the path until it finally recommends a brand.
Can brands actively influence that early, educational part of the funnel, or should a GEO strategy just focus on optimizing for citations at the bottom?
Tom Rudnai: It depends on your brand, but yes, you can absolutely influence it. You just aren't going to show up in a citation during those exploratory conversations.
None of this changes the cognitive process a human goes through to make a complex decision. What I think we'll start to see is that process compress. A buyer will go through that journey over the course of a morning chatting to Claude, rather than over three weeks chatting to sales reps.
As the buyer goes through that process of asking questions, the AI learns about their organization, their needs, and their problems. We've done tests that prove this criteria is retained by the AI and meaningfully impacts the final recommendation it gives.
So, GEO has two goals:
Think of it like an RFP. In sales, there is a cliché that if you receive a blind RFP, you've already lost it because a competitor was in the room helping the buyer write the requirements. If you can produce really high-quality research that influences the AI's collective understanding of a category, you can ensure the AI is setting requirements that naturally funnel buyers toward your product.
Ash Lockett: So how do you create that content? What actually influences an LLM at the top of the funnel?
Tom Rudnai: The most important rule for creating content in an AI world is differentiation rather than optimization. Where search rewarded tightly optimized content, AI rewards differentiation.
The biggest thing we see move the needle is Information Gain. We look at this in three tiers:
The goal is to climb that ladder. Search was a directory; it actively wanted to link out to your website. AI has absolutely no interest in linking out to a brand website. It wants to serve the user directly. If you want it to cite your content, you have to give it a really clear reason why. It has to teach the machine something it doesn't already know.
Ash Lockett: That completely changes content strategy. In an SEO world, you produce as much volume as humanly possible to rank for keywords. Now, fewer, higher-quality pieces that add Information Gain are what matter.
Tom Rudnai: Exactly. We live in an incredibly noisy world where every brand can use AI to optimize their content. When you see "AI slop," it’s actually well-optimized content—it's just bland and unoriginal. The art of marketing right now is intelligently un-optimizing your content by ensuring there is a foundational, human insight at the heart of it.
Ash Lockett: We know that an LLM's understanding of a brand isn't just based on a single piece of new content; it looks at your entire digital estate. A lot of marketing managers have blogs going back four years that articulate their brand entirely differently than they do today. Does that confuse the AI?
Tom Rudnai: Yes. We call it Content Debt. Every new piece of content you publish accrues a little bit of debt, and your maintenance bill goes up.
Think about it: On day one, you decide who your brand is and publish content. Six months later, you pivot slightly. A year later, you pivot again. Over time, a massive gap opens up between what your historical content says you do, and what you actually do today.
AI can navigate that journey a little bit, but you are making it very difficult for the machine to understand your canonical truth. Content teams have historically spent 95% of their time publishing new content. That needs to shift much closer to 50/50, where you are actively auditing, updating, and aligning your historical content library so the AI receives a clear, consistent, and differentiated narrative.
Ash Lockett: A big trend right now is prompt tracking—throwing 50 prompts into a tool to see if your brand is recommended. But a B2B user doesn't prompt the same way an average consumer does. How valuable is prompt tracking right now?
Tom Rudnai: It’s difficult. Most AI visibility platforms strip out all the context to simulate a logged-out, anonymous ChatGPT user. That works perfectly if you are trying to sell toothpaste to average consumers.
In B2B, you are trying to reach a buyer who has spent 12 months trying to solve a complex problem. That context is built into their prompts. When we simulated chains of conversations, we found that altering the context completely changed the AI's recommendations.
For example, a generic prompt might recommend HubSpot 40% of the time and Salesforce 60% of the time. But if a buyer adds specific operational context, those numbers completely shift. If you are just tracking anonymous, generic prompts, you are looking at data that is not reflective of your buyer’s actual experience. You need to build highly specific segments and simulate their exact system prompts to get accurate visibility data.
Ash Lockett: Let’s finish up with a few quick-fire questions so our listeners can walk away with some actionable advice. First: What is one widely accepted B2B GEO rule you think our listeners should bravely ignore starting tomorrow?
Tom Rudnai: Anything that seems like a hack. If someone pitches you a quick project—like changing your schema or adding hidden LLM text—run away. It is a discipline, not a quick fix.
Ash Lockett: If a PR or marketing professional wants to be 10% braver tomorrow morning, what's the first thing they should do?
Tom Rudnai: Be human. Talk about how you felt about a thing, not just the thing itself. And please, stop starting your LinkedIn posts with, "I'm delighted to announce."
Ash Lockett: Where do marketing professionals need to be braver when implementing GEO?
Tom Rudnai: Be brave in managing your client's or board's expectations. Stop treating GEO purely as a performance channel. It is a brand investment and a shop front. And be brave enough to recognize that your existing SEO or PR playbook is not the job of GEO. It requires a cross-functional approach.
Ash Lockett: Finally, what is a recent moment of bravery you've celebrated in your own career?
Tom Rudnai: Putting my ego aside. We raised venture money, and I was determined that we had to be a traditional "rocket ship" startup. Over the last six months, I realized I had made mistakes by trying to force that model. I had to burst my own startup bubble, pivot, and lean back into just doing really good work for our clients. It was daunting, but it put the business on a much better trajectory.
Ash Lockett: Thanks, Tom. It’s been a delight to have you join me today. If anyone wants to speak to Tom about GEO and Demand Genius, you can find his links in the show notes below.