The GEO Show, Ep. 6: The GEO Tools Category Can't Get Itself Cited

GEOforge · The GEO Show, Episode 6

Established SEO platforms get cited by AI 26 times more than the GEO tools built to track it. Including GEOforge.

By Paris ChildressGEOforge7 min read

Six stories dropped in the GEO world this week, and the sharpest one should sting anyone selling AI visibility tools, us included: established SEO platforms are getting cited by AI engines 26 times more often than the GEO tools built to track that exact metric.

We cover all of it on Episode 6 of The GEO Show. Here’s the roundup, with the receipts.

ChatGPT search is explicitly personalized before retrieval

OpenAI’s freshly updated Help Center documentation confirms that ChatGPT can rewrite one prompt into multiple searches, a process known as query fan-out, and can draw on a user’s saved memory when constructing those queries.

That second part is the news. It’s not just that ChatGPT fans a prompt out into several searches, it’s that it personalizes those fanned-out queries using that specific user’s conversation history. Two people asking the exact same question can now trigger meaningfully different retrieval paths before the model has even seen a source. For anyone running GEO tests, this closes the door on relying on a single canonical prompt, or even a single canonical set of fan-out queries, to represent how a brand actually gets surfaced. Testing needs to account for buyer personas and contextual variants, because personalization is now happening upstream of retrieval, not just in how the answer gets written.

Information gain is becoming the clearest antidote to AI content sameness

Ahrefs’ August 20 Deep Dive argues that useful content now has to contribute something beyond what already exists: original data, first-hand experience, or a genuinely new perspective. Ahrefs is careful to note that Google hasn’t confirmed its patented information gain score as a live ranking factor.

Confirmed or not, the direction is clear. Synthesized, derivative content is going to keep losing ground, because it has nothing new to offer a model that’s already trained on a thousand versions of the same take. This is a core reason we build GEOforge’s content production on top of BaseForge, a proprietary knowledge base, so what gets published actually carries information gain instead of reading as AI slop. That’s also what wins citations: brands get cited for contributing something a model didn’t already know.

Topical depth beats broad AI visibility for getting a brand actually named

Semrush analyzed 283,215 citation observations across 1,094 ChatGPT categories and found brands were cited and named in 34% of appearances within closely related categories, compared to just 9% in distant ones.

Depth inside a category correlates far more strongly with getting named than spreading thin across many. If your brand competes across several verticals, the data says go deeper in two or three of them before expanding horizontally into adjacent categories. Breadth-first GEO strategies are working against the grain of how these models actually associate a brand with a topic.

GEO has officially entered institutional procurement

South Dakota State University’s public LLM search optimization RFP has an anticipated award date of August 24, an $80,000 annual ceiling, a two-year initial term, and roughly 9,000 pages in scope. Requirements include SEO as the foundation plus AI answer monitoring, Drupal integration, audit logging, US data residency, and SOC 2 Type II or ISO 27001 credentials.

This is a major US institution writing GEO into a governed procurement process, not a marketing line item. That governance layer, security, permissions, integrations, and reporting, goes well beyond prompt tracking, and we expect more institutional and enterprise RFPs to show up with the same kind of assurance requirements attached.

The GEO tools category can’t get itself cited

SEO Rocket measured 6,322 citations across 56 SEO and GEO tools using Ahrefs Brand Radar. Established SEO platforms averaged 131 cited AI answers each. Dedicated GEO tools averaged just 5. Perplexity generated more than a third of all sampled citations, and the study is vendor-produced, US-only, and caps some high performers at 100 citations per engine.

The GEO tools category is more fragmented and more competitive than the SEO tools category ever was, and that makes sense. SEO tools weren’t easy to build a decade ago, and that market coalesced around Semrush and Ahrefs, with Moz picking up meaningful share along the way. Today there are hundreds of mostly fast-built GEO tools, most focused narrowly on AI visibility measurement, and merely entering the category doesn’t make a brand part of AI’s category model. Every GEO tool out there, GEOforge included, has to demonstrate its own AI visibility to prove it belongs in the category it claims to serve. The category is still barely a year to eighteen months old, so this is going to come down to the best-marketed, best-funded, and highest-quality tools rising to the top, the same way it played out in SEO tooling.

AI crawler access is becoming a governance and monetization decision

Cloudflare’s AI Crawl Control now provides crawler-level monitoring, allow/block policies, robots.txt compliance tracking, and a private-beta pay-per-crawl capability, though Cloudflare notes that WAF rules can override crawler policies.

“Should AI bots crawl our site” stopped being a simple yes/no question. There are now dozens of AI crawl bots, and we’ve seen with at least one client a fake bot spoofing OpenAI’s UserBot to pose as a legitimate crawler while actually phishing. That means brands now have to actively analyze crawl logs, decide whether to whitelist or blacklist specific bots by IP, and stay alert for bad actors using a spoofed crawler identity to probe for vulnerabilities. Crawl access is now a genuine governance and monetization decision, not a robots.txt checkbox.

The bottom line

Four moves to make out of this week’s six stories:

• Build GEO tests around personas, not a single canonical prompt. ChatGPT now personalizes query fan-out with saved memory, so two users can see two different retrieval paths.
• Ground content in proprietary knowledge, not synthesis. Information gain is becoming the clearest signal separating citable content from AI slop.
• Go deep in two or three categories before going broad. Closely related categories cited and named brands nearly four times more often than distant ones.
• Prove your own AI visibility before you claim a category. Entering the GEO tools category doesn’t make a brand part of it, and that standard applies to us too.

Sources & method. ChatGPT query fan-out and saved-memory personalization per OpenAI’s updated Help Center documentation. Information gain analysis from Ahrefs’ August 20 Deep Dive. Topical depth and category citation data from Semrush’s analysis of 283,215 citation observations across 1,094 ChatGPT categories. South Dakota State University LLM search optimization RFP, publicly listed procurement filing. GEO tools category visibility data from SEO Rocket’s analysis of 6,322 citations across 56 SEO/GEO tools, measured via Ahrefs Brand Radar; vendor-produced, US-only, capped at 100 citations per engine for some high performers. AI crawler governance and monetization capabilities per Cloudflare’s published AI Crawl Control documentation. Full breakdown and discussion in Episode 6 of The GEO Show.