The GEO Show is a recurring news roundup hosted by Paris Childress, founder of Hop AI and co-founder of GEOforge, covering the latest developments in Generative Engine Optimization (GEO) and AI search. In this debut episode, Childress breaks down a 300% year-over-year surge in ChatGPT referral traffic to B2B sites, G2 data showing 71% of software buyers now use AI chatbots for research, and Pinterest's GEO-driven 20% organic traffic gain. He closes by arguing that GEO is not simply SEO, introducing information gain as the concept that determines whether AI systems cite your content.
ChatGPT referral traffic to B2B websites grew more than 300% year over year, AI chatbots have become the default research tool for most B2B software buyers, and Microsoft just gave manipulative Generative Engine Optimization (GEO) a name and a penalty. Those are the facts. Paris Childress's read on them, in the first episode of The GEO Show, is that GEO is not a rebrand of SEO. It is a distinct discipline built on a new standard: information gain.
The GEO Show is a recurring news breakdown hosted by Paris Childress, founder of Hop AI and co-founder of GEOforge. Each episode works through the top stories of the day in SEO and GEO, with Childress giving his take on what they mean for marketing teams and agencies. He plans to publish as close to daily as he can, tracking how brands get discovered and cited by ChatGPT, Google AI Overviews, Perplexity, and other AI systems.
Demandbase's analysis of more than 11 billion visits across 1,584 customer instances found that ChatGPT-referred visits to B2B websites rose from 645,000 in June 2025 to 2.6 million in June 2026, an increase of more than 300% in a single year. Childress attributes the jump to two shifts: users treating ChatGPT more like a search engine, with real intent to click through to a source, and ChatGPT's interface making citation links more visible and clickable inside the answer itself rather than tucked into a sidebar.
If that trend holds, he argues, GEO starts to look like a genuine performance channel, one measured by traffic delivered, not just visibility earned.
Yes, according to G2's survey of more than 1,000 B2B software buyers: 71% now use AI chatbots for software research, 51% start their research with an AI chatbot more often than with Google, and 69% say an AI chatbot surfaced a vendor they would not otherwise have considered.
Childress calls this a major shift after ten to fifteen years of the same pattern, where buyers searched Google and clicked through review sites like G2's own profile and category pages. For SaaS marketers still treating GEO as a side project to SEO, he says the data is reason enough to reconsider going GEO-first.
Both Google and Microsoft now say yes, and Childress disagrees. OpenAI recently confirmed in ChatGPT's help center that it uses "query fan-out": a single prompt can be rewritten into multiple targeted searches sent to engines like Google and Bing in real time, with follow-up searches issued based on the initial results. Childress's team has been pulling these fan-out queries from Bing Webmaster Tools and building them back into pages that already have citations, to strengthen them further.
Pinterest offered the clearest evidence yet that GEO can be engineered rather than gamed. The company built a system that uses vision language models to predict search queries, generates semantically matched collection pages for those queries, and interlinks them across billions of assets, an approach Childress compares to programmatic SEO with automated internal linking. The result was a reported 20% increase in organic traffic. As Childress puts it:
This is one of the stronger examples of GEO behaving like an engineering or a search discipline rather than a citation hacking exercise.
Meanwhile, Bing's webmaster guidelines now explicitly name "manipulative GEO" and warn that abusive tactics can lead to reduced rankings, suppressed grounding visibility, loss of grounding eligibility, or delisting, without specifying exactly what counts as manipulative. Both Microsoft and Google frame the fix the same way: do good SEO and good GEO follows.
Childress pushes back. He agrees the two disciplines overlap heavily on-site, in crawlability, site structure, and information architecture, but argues the real differentiator is information gain: whether your content teaches AI crawlers something they have not already ingested from elsewhere on the internet. He positions information gain as the successor to the old EEAT standard (experience, expertise, authoritativeness, trustworthiness), grounded specifically in proprietary data and firsthand expertise rather than the polish of the content itself.
Five stories, one thread: the mechanics of GEO are becoming visible and measurable, and the platforms that used to be gatekeepers are starting to write rules for how it should be done. Childress plans to keep tracking this ground on a near-daily cadence.
Hi, everyone, and welcome to the GEO Show. We're recording on Friday, August 21, and this is the first episode of the GEO Show. I'm Paris Childress. I am the founder of Hop AI, also co-founder of GEOforge, and I'm going to be the primary host of the show. Bear with me in this first episode, because I'm just getting my footing here.
The format for this show is going to be that I read through some top stories of the day as they relate to SEO and GEO, primarily GEO. I've got a really great brief coming in with some top stories, and I'm going to make some comments on how I feel that marketing teams and agencies are impacted by some of these stories, and give my overall response and reaction. That is going to be the predominant format of this show. So let's get started. I have some top headlines I want to run through, and we'll see what this all means.
The first comes from Demandbase, and the headline is that ChatGPT referral traffic to B2B sites is accelerating sharply. It's confirmed by Demandbase, who analyzed more than 11 billion visits across 1,584 customer instances, and found that ChatGPT-referred visits rose from 645,000 in June of last year to 2,600,000 in June of this year, 2026. That's an increase of over 300% year over year. This is major, and I think what this is telling me is that the referral traffic game in GEO is really changing. That's a 3x increase in just one year for ChatGPT alone.
I do think that it's a reflection of maybe a couple of factors. Number one is that users are starting to click through more often, and maybe that's because they're starting to substitute ChatGPT for Google and use it more in the mindset of a search engine, where they intend to get an answer and then pursue that answer deeper through a click. That's one factor: the user behavior side. The other factor, I think, is that ChatGPT, in its user interface, is probably making changes that make links more visible within the answers themselves, and that's something I've been noticing too. If citations are appearing as clickable links within the answer, I think those are going to get a higher click-through rate than citations that appear in a sidebar, where users would have to take their attention away from the answer to view and scroll through separately.
That's really interesting, and I hope that trend continues, because it's going to more and more position GEO as a performance channel the way SEO has always been positioned as a performance channel, meaning it's responsible in part for driving actual traffic to a website. Alright, let's move on to the next topic, and that's coming from G2. The headline here is that AI chatbots are now the number one shortlist influence for many software buyers. G2 confirmed, through a survey of over a thousand B2B software buyers, that 71% of respondents use AI chatbots for software research.
51% start with AI chatbots more often than Google, and 69% said AI surfaced information that led them to choose a different vendor than expected. This is also a major shift in the way software buyers are conducting their journey. The classic journey for B2B SaaS, for at least the last ten to fifteen years, has been that you go on Google, typically search for a keyword, and get the results. Many times G2 and their software profile pages and category pages will rank competitively in Google, so you may find yourself clicking through G2, but that search was the dominant research journey, the starting point for the dominant share of B2B software-buying journeys. Now it's shifted in a major way to chatbots. So 71% are using AI chatbots for software research.
This has major implications for SaaS marketers. I would say for those SaaS marketers who are doing SEO but not yet fully bought in and fully invested in GEO, I think it's time to really reconsider going GEO-first due to that fact. It doesn't really surprise me that much, actually. Alright, let's move on to the next story.
OpenAI, in its help center, is now explicitly stating that it's documenting something called search query fan-out. What search query fan-out is, is that it takes the user's prompt and repurposes it into a series of search queries that it then takes into search engines like Google and Bing in real time. This process is called query fan-out, and ChatGPT has said publicly, in its help center, that it may rewrite one user request into one or more targeted searches, inspect initial results, and then issue additional searches to other providers. So they are really now forecasting, and admitting to, something I think the industry has known for a while: when real-time search happens in LLMs, meaning an LLM cannot immediately generate a predictive reply from its pre-training dataset, it goes out and does a real-time search, or in fact does several real-time searches, using this query fan-out method.
One of the tactics we've been exploring here in our agency is to go through Bing Webmaster Tools, which does surface the query fan-out queries by page, and to look at those queries, then figure out how we can take them and build them back into the pages that have already gotten citations, to further strengthen them. That's one interesting tactic, and I'd love to hear feedback if anybody else is doing something similar. Alright, moving on to the next story.
Let's talk about Pinterest. This is kind of a GEO/SEO story. Pinterest has published rare production evidence that GEO can drive actual organic growth. Most people have been talking about zero-click, and that GEO is kind of living in a world of zero-click, meaning GEO works not because it drives traffic, but because it makes your brand visible and causes people to then start looking for you in other places. But Pinterest has confirmed, through research, a deployed GEO system that uses vision language models to predict search queries, and then, with that, creates semantically coherent collection pages and builds authority-aware internal links across billions of assets.
Let's break that down and repeat it. The first thing they do is use vision language models that predict search queries. I presume these are going to be GEO search queries, or Google search queries that may pull up AI Overviews. What they're doing is, using that model of predicted queries, creating Pinterest collection pages that match.
I guess they're kind of creating landing pages semantically based on those search queries, and then interlinking all of those collection pages. That sounds like programmatic SEO with internal linking to me. The result was 20% organic traffic growth, which is really major for Pinterest, a multi-million monthly active user impact. This is one of the stronger examples of GEO behaving like an engineering or a search discipline rather than a citation hacking exercise.
I found that really interesting, and I think this is going to be a tactic people start using as well, through what we discussed earlier with that query fan-out approach, where you can take a set of query fan-outs and use that to try to predict the type of page you're missing, then do that at scale. But, of course, Pinterest is one of the rare companies that has the ability to do this at such a massive scale. Alright, moving on to what I think is going to be the last story of the day, and that's from Microsoft Bing. It says Microsoft is drawing an explicit policy line around manipulative GEO.
This is the first time I have come across that term, manipulative GEO. We've heard manipulative SEO and black hat SEO for maybe the better part of two decades now, but this is the first time I've really come across a publication like Microsoft using this phrase. It's acknowledging the fact, actually, that GEO can be manipulated. Bing's webmaster guidelines now explicitly discuss Generative Engine Optimization, GEO, and state that abusive tactics can lead to reduced rankings, suppressed grounding visibility, loss of grounding eligibility, or delisting. Interesting.
Bing says traditional crawlability, consolidation, clarity, structure, and trust signals support both SEO and AI grounding. That sounds very similar and very familiar to what Google stated officially about a month ago, which effectively said that GEO is SEO, meaning that if you want GEO results, then just keep doing good SEO. It's interesting that Microsoft has now followed suit, and maybe taken it even a step further, by acknowledging, by stating this term manipulative GEO, and then saying there are penalties involved with manipulative GEO, without actually saying what manipulative GEO is. Is it scaled AI-generated content? Is it scaled citation building? Is it something else? We're not really sure yet, because we do know manipulative SEO has to do with scaled low-quality content and automated link practices, for example.
I find this very interesting, but I do tend to disagree with both Google and now Microsoft in their implication that GEO is just SEO. I think it's very different, while, of course, there are overlapping aspects to both. I think the biggest area of overlap between SEO and GEO is on-site technical factors. You still need a website that performs really well, that's easy to crawl, with content that's very accessible by all sorts of crawlers, and a well-organized, logical structure and hierarchy of information. Information architecture is key there.
That's the biggest overlap. The other overlap I see between SEO and GEO is content quality and content uniqueness. The acronym EEAT, which stands for experience, expertise, authoritativeness, and trustworthiness, I believe was the gold standard for SEO content, and now I believe that has evolved into a concept called information gain, which we strongly subscribe to at GEOforge.
Information gain is basically the degree to which your content can provide new learning material, new training material, to LLMs and to AI crawlers. Basically, can you teach AI crawlers new information by providing it with novel data and proprietary knowledge it has not already crawled and ingested across the internet? That concept is information gain. I think content with high information gain, grounded solidly in some sort of knowledge base, is also very similar to the older concept of EEAT-friendly content, because it naturally brings out a person's, or a company's, or a brand's unique experiences, perspective, and expertise on a subject matter. So I guess Microsoft has now put its flag in the ground along with Google, saying: we don't want anybody doing manipulative GEO.
You should all just be doing SEO as you have been doing for many years, and you'll get good GEO results as a result. Alright, I think that's going to wrap up this session. This was the first episode of the GEO Show. I'm going to try to do this as often as possible. I'm going to aim for daily, and I'm going to try to keep bringing you every day the top stories, the top news within the GEO space, with my own commentary.
I'm looking forward to hopefully building an audience behind this, and getting your involvement and your feedback. Thanks for listening, and until next time.
