We build and maintain websites for enterprise clients across WordPress, Shopify, custom applications, and headless platforms. Most of the content those sites produce is getting cited in AI answers right now. Whether the brand name shows up in those answers is a different question entirely, and for most organizations, the answer is not as often as it should be.
A February 2026 Semrush study analyzing nearly 4,000 domain appearances across four major AI platforms found that 62% of AI citations are what researcher Kevin Indig calls ghost citations: the AI uses your page as a source, links to it in the footnotes, and never once mentions your brand name in the answer. The user gets the information. They have no idea who produced it. Your competitor, whose brand the AI already associated with the category, gets the recommendation.
Organizations with strong informational content libraries get cited regularly and named rarely. The gap isn't about content quality, instead it's about brand authority, and those are different problems with different fixes.
Why Content-Heavy Sites Are Particularly Exposed
A significant share of the informational content on the internet lives on content management systems, WordPress, Shopify blogs, headless CMS platforms, custom builds. Blog posts, resource libraries, how-to guides, industry explainers, the content types these platforms produce most naturally are exactly the content types most susceptible to ghost citations.
Semrush's data shows that informational queries, the "what is," "how does," and "explain" searches, produce an 89.3% citation rate but only an 18% mention rate. Comparative queries, the "best," "vs," and "recommend" searches, produce a 43.3% mention rate. The AI names brands when it's comparing options. When it's explaining something, it extracts the information and leaves the author behind.
Most content strategies, regardless of platform, are built around informational content. It's where organic search volume lives, it's where EEAT signals accumulate, and it's where content-driven sites have historically competed well.
This isn't an argument to stop producing informational content, but an argument to produce it differently.
What's Actually Happening Inside the AI
The leading explanation for why ghost citations happen, supported by behavioral analysis across 541,213 LLM responses by Seer Interactive, is that citations are added after the answer is already written. The AI decides which brands to recommend first, drawing on its training data. Then it goes looking for sources to support those choices.
Your content passes the retrieval check. Your brand doesn't pass the recall check. Those happen at different points in the process, and improving one doesn't automatically improve the other.
The data signature that supports this: when a brand is mentioned in an AI response, its citation rate is 53%. When the brand is not mentioned, that same brand's citation rate drops to 10.6%. A 5x difference. If retrieval were driving recommendations, those numbers would be similar. They're not.
What this means practically: a site that publishes excellent, well-optimized content can earn citations across dozens of AI queries while the brand behind it remains effectively invisible in every answer. The content is doing its job. The brand isn't getting credit for it.
The Content Structure Problem
There's a specific structural issue that makes most CMS-published content particularly easy to ghost-cite. Most blog posts, whether published on WordPress, a Shopify blog, or a headless CMS, are written to rank for a query, not to attribute a claim to a named organization.
When we audit a site for ghost citation exposure, this is the first thing we check: can the AI lift the insight without lifting the brand name? On most sites, the answer is yes. The fix isn't complicated once you know what to look for, but it requires deliberately restructuring how claims are attributed in the body copy.
"There are five approaches to WordPress security" is a ghost-citable sentence. "Curious Minds has found across our enterprise client work that WordPress security comes down to five recurring failure points" is not. The information is the same. The second version makes the brand name grammatically necessary to extract the claim.
This applies across any content type: case studies, how-to guides, security explainers, platform comparisons. If the insight can be lifted without the name, it will be.
Why Strong Brands Don't Have This Problem
The Seer Interactive analysis found that category owners, brands whose names are effectively synonymous with a space, have ghost citation rates approaching zero. Industrial services clients in the study showed rates of 0.3%. Financial services and HR technology both came in under 2%.
These brands got there through years of consistent investment that made their names the default answer in their categories. The AI learned the same associations the market already held.
The Semrush data showed the same pattern at the domain level. Google was named in AI answers nearly three times more often than it appeared as a source link. Medium.com was cited 16 times and never named once. Aggregator sites get used as reference material. Brands with strong identities get named.
For agencies, professional services firms, and B2B organizations building content programs, this is the clearest signal in the data: content volume doesn't close the ghost citation gap. Brand authority does. A site that publishes more informational articles without addressing the underlying brand recognition problem will widen the gap, not close it.
What We Do About It
The fix operates at three levels, and none of them produce results overnight. AI systems update on training cycles, not in real time. Changes made today are investments in how the next model version perceives your brand.
The first level is structural, making brand attribution impossible to extract from. This means rewriting key content so that claims are grammatically attached to the brand name, not floating free of it. For most sites, this is an audit and rewrite of the highest-cited pages rather than a content overhaul.
The second level is entity infrastructure, the machine-readable signals that tell AI systems who you are and what category you belong to. Organization schema with sameAs markup. Consistent canonical brand name across all properties. Author schema connecting named individuals to the organization. FAQ schema where the brand name appears inside the answer text, not just the question. These are the signals the model reads when deciding which brands to name. If they're absent or inconsistent, the model defaults to whatever brand it has seen most frequently in recommendation contexts.
The third level is off-site brand reinforcement, earning mentions of your brand name in recommendation contexts on authoritative third-party domains. Industry publications, analyst coverage, partner pages, professional directories. The AI learned your competitors' names from somewhere. Building the same signal for your brand is a PR and content distribution problem as much as an SEO one.
Across the platforms we work on, the combination of on-page attribution restructuring and entity schema implementation is usually the highest-leverage starting point. The off-site work takes longer and involves more stakeholders, but the on-page and technical fixes can be scoped and executed within an existing content or development engagement.
What to Track
Ghost citations require different measurement than traditional SEO. Session counts and rankings don't surface them.
Track these three things for AI visibility:
Branded search volume trends
AI citation tracking across ChatGPT, Perplexity, and Google AI Overviews segmented by query type
Multi-touch attribution analysis for actual leads
The Semrush data also points to query phrasing as a variable worth tracking separately. Short conversational queries produce 30 to 50 times more brand mentions than long structured prompts on the same topic. Understanding which query types your content is being retrieved for and whether those queries produce mentions or ghost citations shapes which content investments are worth making next.
Where This Is Going
The ghost citation problem is likely to get more pronounced before it gets better. AI systems are processing more queries, citing more sources, and the gap between brands that have built parametric recognition and those that haven't is compounding with every model training cycle. Organizations that address the brand entity problem now are building an advantage that compounds over time.
Organizations that continue producing informational content without addressing the underlying recognition gap are widening the distance between their content's reach and their brand's visibility. The window to close that gap cost-effectively is now, before the category associations in the next generation of AI models are set.
How Curious Minds Can Help
This work sits at the intersection of SEO, content strategy, and technical implementation, which is where our practice lives. We audit content ecosystems across WordPress, Shopify, headless, and custom platforms for ghost citation exposure, identify the highest-cited pages where brand attribution is absent, implement entity schema that gives AI systems the signals they need to connect your brand name to your category, and adjust content structure so that the AI can't extract the insight without the name.
This is newer work and we're honest about what's confirmed versus what's still being measured. What the data makes clear is that adding more content without addressing brand entity recognition will not fix a ghost citation problem. For sites with established content programs, the leverage is usually in optimizing what exists rather than producing more.
If your content is being cited in AI answers but your brand isn't showing up in the responses, that gap is worth understanding and worth closing before it compounds further.
