AI Citation Share Is Deserting Aggregator Sites, and the Gap Is Bigger Than Anyone Budgeted For

If you run marketing for an enterprise brand and you've been quietly assuming your AI search citation share tracks your Google rankings, stop. New data says it doesn't, and the gap is not small. Aggregator sites — marketplaces, comparison pages, review directories — control 76.0% of the SEO visibility split against primary and direct brand sites. On ChatGPT, that same group of aggregators holds only 38.3% of the citation pool. Same companies, same two buckets, two totally different outcomes depending on which system is doing the ranking.

The AI citation share number that should reset your GEO budget

Every enterprise SEO team has some version of the same mental model: get the page ranking, and the AI citation follows eventually, because AI engines are supposedly just skimming the same organic results everyone else fights over. That model was always a simplification. This data makes it look wrong in a specific, measurable direction. Aggregator sites built their SEO dominance the old way — comprehensive category coverage, huge internal linking graphs, decades of domain authority compounding on comparison and directory pages that rank for thousands of long-tail terms at once. That machine still works on Google. It built 76.0% of the visibility split.

ChatGPT is not running that machine. When it decides who to cite, the same aggregator advantage collapses to 38.3% — barely more than a third of the citation pool, against primary and direct brand sites taking the other 61.7%. That's a 37.7-point swing between the two systems for the exact same set of companies. If your GEO plan assumes AI citation share is just downstream of AI search visibility, this is the single stat that should make you re-check the assumption before you allocate another quarter of budget on it.

Read the bars the other way and the story gets sharper. Primary and direct brand sites — the manufacturers, SaaS vendors, and service providers that comparison sites write about, rather than the comparison sites themselves — hold only 24.0% of SEO visibility but 61.7% of ChatGPT's citations. A brand site that's invisible in the top ten Google results for a category term can still be the thing ChatGPT names when a user asks for a recommendation. That's not a footnote. For a lot of enterprise marketing organizations, it's the whole strategic picture inverted.

What counts as an aggregator here

Indig's split draws a clean line: aggregator sites are marketplaces, comparison and review sites, and directories — the pages built to compile, rank, and compare other companies' offerings rather than sell or deliver their own. Primary and direct sites are everyone else in the tracked set: the brand's own domain, where the product, service, or company being discussed actually lives. It's the same distinction Something Inc. has made in prior coverage of comparison pages winning citations and in the broader anatomy of an AI citation — a citation from a third-party comparison page and a citation that points straight at the brand's own site are structurally different assets, and they behave differently depending on which engine is doing the citing.

This distinction matters because it's easy to conflate "getting mentioned by a comparison site" with "getting cited directly." Both are real GEO outcomes and both count toward visibility in a loose sense. But the 76% versus 38.3% gap is specifically about which type of domain earns the citation link itself — whose URL shows up in the source list when ChatGPT answers a query. An aggregator ranking you favorably inside its own comparison page doesn't help if ChatGPT decides to cite a different comparison page, or skips the category page altogether and cites the vendor directly.

Why AI citation share diverges from SEO visibility

This is Something Inc.'s read on the mechanism, not something the underlying research claims to explain, but the shape of the gap points somewhere specific. Google's ranking algorithm rewards comprehensiveness and link-graph authority at the category level. A comparison page covering forty vendors, updated regularly, linked to from a hundred other pages on the same domain, is exactly what two decades of Google ranking signals were built to reward. It doesn't need to be the authoritative source on any single vendor — it needs to be the most useful single page for someone comparing all of them, and that's a genuinely different, genuinely valuable thing to build.

ChatGPT isn't ranking pages for a results list. It's assembling an answer, and an answer favors a source that is the thing being described over a source that lists the thing being described. Ask ChatGPT what a specific product costs, what its refund policy is, or how its onboarding works, and a comparison page's secondhand summary is a worse source than the vendor's own pricing or policy page — thinner on the specific facts being asked for, more likely to be stale, and one layer removed from the primary information. A generation system built to ground its answer in the most directly relevant source has a real reason to prefer the primary domain, not just a bias toward it.

There's a second factor worth naming plainly: many aggregator pages exist to be broad rather than deep. A directory listing covering hundreds of companies in a paragraph each has thin, generic language per entry — exactly the kind of content that scores poorly against the six link types AI engines trust, most of which reward specificity and original information over compiled summaries. Google can reward that page for its aggregate usefulness across an entire category. An AI engine answering one specific question about one specific company has no comparable incentive to route through the thin paragraph when the deep page it's summarizing is one click away and directly citable.

What this means if you run a primary/direct brand site

If your brand's own domain is the primary/direct site in this split, the data is good news with a catch. You're starting from a structural advantage on the AI citation side that you likely aren't fully capturing yet, because most enterprise SEO programs are still built around the assumption that aggregator-style comprehensiveness is the thing to chase. The 61.7% figure isn't a guarantee, it's a ceiling that primary sites as a category are closer to than aggregators, and closing the gap between where your brand sits today and that ceiling is a content and structure problem, not a link-building one.

Concretely, that means treating your own pricing pages, product documentation, comparison-to-competitor pages, and case studies as the primary citation surface, rather than assuming a category-level comparison site will do that work for you. Something Inc.'s enterprise GEO readiness framework covers this in more depth, but the short version tracks directly with this data: a primary site earns its 61.7%-class share by being the most complete, current, and specific source about itself, not by trying to out-aggregate the aggregators at their own game. That's a fight primary sites structurally lose on Google and don't need to fight on ChatGPT.

It also changes where reporting attention should go. A brand tracking only organic rank and share-of-voice against a category page will keep missing the fact that its own domain may already be winning the more consequential battle — the one where an AI answer names the brand directly instead of routing a user through a comparison page first. Something Inc.'s reporting and analytics approach for GEO clients splits these two metrics apart for exactly this reason: aggregate SEO visibility and AI citation share move independently enough now that blending them into one dashboard number hides the story this data is telling.

What this means if you run an aggregator

If your business model is the comparison page, the marketplace, or the directory, don't read this as a death notice for the category. Aggregators still hold 76.0% of SEO visibility, and that traffic and its associated revenue aren't evaporating. What's changing is the second channel sitting next to it, and 38.3% versus 61.7% is a real gap to be losing when AI-driven referral and answer-engine traffic is growing as a share of total discovery. An aggregator ignoring this split is optimizing for a channel that's flat to declining relative to one that's structurally underweighted for the category as a whole.

The fix isn't abandoning the comprehensiveness that built the SEO share — it's adding depth on the specific comparisons ChatGPT is more likely to ground an answer in. A category page that expands its most-compared pairs into genuinely deep, current, fact-dense sections — pricing detail, feature-by-feature specifics, dated update timestamps — starts to look more like a primary source on that narrow comparison, even while remaining an aggregator at the domain level. Something Inc.'s third-party AI citation playbook found depth on the specific pairing being compared, not just breadth across the category, is what separates aggregator pages that do get cited from the ones that don't. That's the lever available inside the current 38.3%, and it's a content investment, not a rebuild.

It's also worth checking whether the citation gap is concentrated in a specific vertical rather than spread evenly. The underlying index covers 26 verticals, and a category with fast-moving pricing or frequent feature changes — B2B SaaS is a good example — punishes stale aggregator content harder than a slower-moving category does, because the freshness gap between the comparison page and the vendor's own site widens faster. An aggregator in a fast-moving vertical has more urgency here than one in a category where facts don't change much month to month.

The math nobody wants to hear

Here's an illustrative scenario, not sourced data: imagine a $50M-revenue SaaS category where AI-assisted research already influences a meaningful share of vendor shortlists before a sales conversation ever starts. If aggregator sites are capturing only 38.3% of the citations that shape those shortlists, relative to their 76.0% share of organic visibility, an aggregator business built around SEO-driven traffic is over-indexed on a channel that's already ceding ground on a second, growing one. That's a hypothetical illustration of the stakes, not a projection — the actual dollar impact depends entirely on how much research activity has already shifted from search results to AI answers in a given category, which Something Inc. isn't claiming to quantify here.

What is not hypothetical is the direction of the gap and its size. A 37.7-point swing between two ranking systems, measured on the same roughly 2,600 companies across 26 verticals, isn't noise or a methodology artifact — it's a structural difference in what each system is built to reward. Google rewards the page that lists the answer. ChatGPT rewards the page that is the answer. Related work on this site — the 2026 AI citation study and the ownership model for AI visibility — has been circling this same distinction from different angles for months. This data puts a number on it.

One more thing worth stating plainly: this is a single data point from a single index, covering one fetch date for AI mentions and roughly four weeks of SEO visibility. It's a strong signal, not a longitudinal trend line, and Indig's own SSI framework is designed to be tracked over time rather than read as a one-time snapshot. Treat the 76.0% and 38.3% figures as the starting measurement for a gap worth watching quarter over quarter, particularly for any brand or aggregator making a budget decision based on it today.

Do this next

Pull your own citation data this week and split it the same way Indig split his index: which of your cited sources are primary/direct pages you or a partner brand controls, and which are aggregator pages listing you alongside competitors. If your ratio looks nothing like 61.7/38.3 in the direction that favors primary sources, that's not proof something's wrong — categories vary — but it's a reason to look harder at whether your own domain's pricing, comparison, and product pages are deep and current enough to be the source ChatGPT reaches for instead of the directory listing you don't control. If you need a structured way to run that audit, Something Inc.'s generative engine optimization team builds this exact primary-versus-aggregator citation breakdown as the first deliverable on every new engagement, precisely because the gap this data describes is invisible in a standard SEO dashboard.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.