Your Website Is Not Where AI Gets Its Information About You
A study of 37,547 AI citations reveals an uncomfortable truth: only 5% point to a brand’s own site. Here’s what that means for every marketing leader — and how planeed opens a fundamentally new path forward.
Most marketing teams have spent years building their website into the best possible version of their brand. Clear positioning. Strong content. Consistent messaging. Optimized for search. And it is good, necessary work.
But there is a question worth sitting with: when ChatGPT, Perplexity or Google AI Overviews answer a question about your brand, how much of what they say comes from your website?
A research study published in June 2026 by GEO analytics platform llemmy provides the answer. It is not the one most marketing leaders are expecting.
The Number That Changes Everything
The study analyzed 37,547 citations across 5,200 AI-generated answers from ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews — spanning three brands across three very different industry categories. They tracked every source cited in every answer and categorized where it came from.
The headline finding:
5 %
of all citations pointed to the brand’s own website.
95 %
pointed to third-party sources.
That means roughly 19 out of every 20 citations that AI systems generate when answering questions about a brand come from somewhere other than that brand’s own pages. The content you wrote, the pages you optimized, the messaging you refined – it accounts for a rounding error in how AI systems form and express their understanding of who you are.
AI doesn’t repeat what you say about yourself. It repeats what others say about you. For marketing leaders, this is one of the most important shifts in the last decade.
Where AI Actually Gets Its Information
If only 5% comes from brand-owned domains, where is the other 95% coming from? The breakdown reveals a clear pattern:
- 56% came from a long tail of category-specific sources – comparison sites, niche directories and vertical publishers specific to each industry.
- 15% came from review sites and directories – Yelp-style listings, comparison platforms and “best of” aggregators.
- 14% came from social platforms – Facebook, Instagram and similar channels.
- 4% came from forums and user-generated content – most notably Reddit, which outperformed every traditional news outlet.
- 2% came from news and editorial sources.
- 5% came from brand-owned sites.
The picture this paints is striking. AI systems are not reading your brand through your own communication. They are reading it through the ecosystem of sources that exist around you – and forming their picture of your authority, credibility and relevance from what that ecosystem says.
What This Means for Your Marketing Strategy
The research has immediate and concrete implications for how marketing leaders should allocate attention, budget, and effort:
Your website is necessary but not sufficient.
On-site optimization addresses roughly 5% of where citations come from.
Third-party presence is your primary AI visibility lever.
Review listings, directory profiles, social channels, niche publishers – these are the sources AI actually quotes.
Category-specific sources require category-specific research.
The 56% long tail is different for every industry. Identifying which specific sources AI draws from in your space is the highest-value work in GEO strategy.
Consistency across all sources is non-negotiable.
If your brand is described differently across channels, AI receives a fragmented signal. Fragmented signals produce ambiguous, inaccurate or absent AI representation.
The Problem No One Has Solved — Until Now
Here is the challenge that the data exposes – and that most of the GEO industry quietly sidesteps.
You know AI is reading third-party sources. You know your social profiles, directory listings, review platforms and category-specific publishers are where the citations actually come from. But you cannot place your schema markup there. You do not own those websites. You cannot embed your structured data in a trade magazine article, a LinkedIn post on someone else’s page or a comparison platform you have no control over.
This is the technical impossibility at the heart of AI visibility work. The largest share of what ChatGPT and Perplexity cite comes from third-party media – but a company cannot embed its own semantic code on sites it does not own. Classic SEO/PR and standard GEO tools can only optimize what lives on your own domain. Which, as the data shows, is a rounding error.
The gap between where AI citations come from and where brands can currently apply structured data is the single biggest unsolved problem in AI visibility. planeed is the first platform to close it.
What AI Says About planeed’s Approach
We asked AI to evaluate planeed’s architecture – specifically, the model of hosting a company’s knowledge graph on planeed.org (functioning as your brand’s official record in the semantic web) and making it linkable from third-party media. The response was unambiguous:
These are not marketing claims. These are the words of one of the AI systems that determines brand visibility – assessing planeed’s approach against the full landscape of existing solutions.
How planeed Compares to Every Other Approach
Google’s AI laid out a direct comparison of methods. The effectiveness ratings speak for themselves:
Method
Effectiveness for LLMs
Classic SEO / PR
Medium. AI understands the text but must interpret it (risk of hallucination).
Standard GEO
Medium to Good. Only helps if the AI uses your website directly as a source.
The planeed Approach
Excellent. Provides AI with verified facts structured across domains, sustainably minimizing hallucinations.
Classic SEO and standard GEO are rated Medium because they only reach the sources a brand already owns. The planeed approach is rated Excellent because it reaches the sources AI actually cites – the 95% that live outside the brand’s own domain.
The Three Things planeed Unlocks That No Other Tool Can
Google’s AI identified three specific advantages of planeed’s architecture that no other solution currently offers:
The Technological Data Bridge to Third-Party Media
The market problem: Over 85% of AI citations come from third-party sites — trade magazines, news portals, LinkedIn, directories. But companies cannot embed their own schema code on sites they don’t own.
What planeed unlocks: planeed solves this technical impossibility. By hosting the brand’s knowledge graph on planeed.org and making it linkable from third-party media via semantic ‘bridge links’, external PR and editorial coverage can be directly connected to a brand’s entity graph. No other tool on the market can currently do this.
From Marketing Speak to Machine Fact (Grounding-as-a-Service)
The market problem: AI systems ignore classic marketing copy because they are looking for verifiable facts to avoid hallucinations. Unstructured corporate data – product descriptions, leadership bios, capability claims – is largely invisible to AI systems at the entity level.
What planeed unlocks: planeed translates unstructured corporate data into a logical graph structure (DefinedTerms and Relations), delivered to AI crawlers in the exact format they require. For the AI, the trustworthiness of the brand rises toward 100% because the facts are verified by the network, not just asserted on a single page.
Zero IT Load for the Marketing Team
The market problem: Large corporate websites have rigid CMS structures. Any change to core schema architecture requires time and technical understanding. Most marketing teams do not have this ressources.
What planeed unlocks: Because the knowledge graph is hosted independently on planeed.org, the internal IT effort is completely eliminated. Marketing can control, expand and maintain the brand’s Semantic Brand Architecture entirely autonomously – without opening a single IT ticket.
The Bidirectional Validation Effect
There is a mechanism in planeed’s architecture that Google’s AI specifically highlighted as particularly powerful — what it called “Bidirectional Validation” or the “Two-Way Proof.”
Here is how it works: when an AI crawler reads a third-party article (a trade magazine piece, a partner publication, an industry directory entry) and encounters a link pointing directly to planeed.org/entity/your-brand, the loop is closed. The AI no longer just sees the information on your website. It receives an independent, machine-readable confirmation from a third-party source.
For an AI’s trust score, this is the strongest possible signal. It is the equivalent of having multiple credible, independent witnesses confirm the same fact – rather than relying solely on what you say about yourself.
This is the mechanism that makes planeed’s approach qualitatively different from every other AI visibility tool. It is not just optimizing what the AI finds on your own pages. It is building a network of verified, cross-domain signals that give AI systems the corroboration they need to represent your brand with confidence.
Every Marketing Leader Should Ask Now
The research provides a useful diagnostic lens. Here are three questions worth taking to your next strategy conversation:
- Where does AI currently cite our brand? Not on your own website – in the third-party sources your category’s AI answers actually draw from. Do you know which directories, comparison platforms, social channels and niche publishers the engines cite when answering questions in your space?
- How consistently is our brand described across those sources? Check how your brand is described in each of the third-party sources AI systems cite in your category. Is the positioning consistent? Are your products and expertise areas described the same way across every source – or has each developed its own version?
- What would an AI system conclude about our brand from the ecosystem around us? If you aggregate the signals from your review listings, social profiles, directory entries, community mentions and third-party coverage, what picture emerges? Is it the picture you intend – or a fragmented, outdated or competitor-adjacent version of it?
The Strategic Shift This Requires
For two decades, digital marketing strategy was oriented around owned media. Your website, your content, your channels – the things you controlled were the primary levers of visibility.
AI search shifts the center of gravity toward earned and third-party presence. And planeed is the first platform built specifically to bridge that gap – to connect what a brand owns with the ecosystem it does not own, in a way that AI systems can read, verify and trust.
The data is clear. 95% of AI citations come from outside your website. The marketing leaders who understand this – and who build the semantic architecture necessary to reach that 95% – are the ones whose brands will be recommended.
95% of AI citations come from outside your website. planeed is the first platform to build a data bridge into that 95%. That is not an incremental improvement over existing GEO tools. It is a different category entirely.
