Machine Relations

The Entity Concentration Crisis: Why 59.5% of AI Citability Is Going to 10 Experts

Among 143 digital-marketing experts Authoritas tracked, the top 10 captured 59.5% of weighted AI citability in February 2026, up from 30.9% in December. What it means for brands.

Jaxon Parrott
Jaxon ParrottFeb 21, 2026

In December 2025, the top 10 of the 143 digital-marketing experts Authoritas tracks captured 30.9% of weighted citability across ChatGPT, Gemini and Perplexity. By Authoritas's 2 February 2026 measurement that number was 59.5% — a 92% increase in under two months. The Herfindahl-Hirschman Index — the standard measure of market concentration — for that expert pool rose 293% in the same window, from 0.026 to 0.104. It is not a slow shift. In this dataset it is a move toward winner-take-most.

This is what we at AuthorityTech have been tracking under the discipline of Machine Relations — the practice of earning AI engine citations and recommendations for a brand. The entity concentration crisis is the most urgent inflection point in that discipline. Every day a brand waits to optimize their entity signals, more of the available citation share gets locked up by competitors who moved first.

Here's what the data actually shows, why it's happening faster than anyone predicted, and exactly how brands can still claim entity authority before the window closes.

Key Takeaways

  • The top 10 of 143 tracked digital-marketing experts captured 59.5% of weighted AI citability in February 2026, up from 30.9% in December 2025 — a 92% concentration increase in under two months, per Search Engine Land's February 2026 analysis of Authoritas data. The score is weighted across ChatGPT, Gemini and Perplexity.
  • The Herfindahl-Hirschman Index of that 143-expert pool rose 293% from December to February 2026, from 0.026 to 0.104, a sharp move toward winner-take-most in that pool.
  • 92-94% of Google AI Mode sessions ended without a visit to an external site in Semrush's clickstream study of US desktop sessions (May to July 2025) — so inside an AI Mode answer, being cited is often the only visibility a source gets.
  • The knowledge graph market is forecast to reach $4.46 billion in 2030, growing at a 22.5% CAGR, reflecting the infrastructure bet being made on entity-centric AI systems.
  • The Search Engine Land article's author argues that entities present in all three AI graph layers — entity graph, document graph and concept graph — receive a disproportionate, multiplicative boost over single-layer competitors, per his three-graphs framework. It is an explanatory model, not a measured result.

What Is the Entity Concentration Crisis?

AI search engines don't browse the web the way Google crawled it. They resolve entities — structured representations of who you are, what you do, and what you're authoritative about. When ChatGPT, Perplexity, or Gemini answers a question about your category, it's pulling from a knowledge graph where entities that are well-defined, consistently represented, and broadly corroborated get cited. Entities that are ambiguous, inconsistent, or poorly corroborated get skipped — regardless of how much content you've published or how high you rank on Google.

The crisis is this: in the Authoritas dataset, citability is compressing quickly toward the experts who actively manage their entity signals, per the Search Engine Land article's reading. The 293% jump in that pool's HHI shows how fast a top tier can pull away. In this dataset the concentration rose over two months, not decades.

Once a handful of entities dominate citation share in a category, the flywheel becomes self-reinforcing. AI systems learn from their own outputs. The more a brand gets cited, the more training signal confirms it as the authoritative entity. The more it gets cited again. Early movers aren't just winning today — they're training tomorrow's models to prefer them.

The Three-Graph Model: Where Entity Authority Lives

One explanation for why concentration happens is how AI knowledge systems are structured. The author of the Search Engine Land article that reported the Authoritas data argues that entities present across what he calls the "three graphs" — knowledge graph, document index and concept space — receive a disproportionate, multiplicative boost in retrieval over entities present in only one. That is the author's framework, not a measured result.

The three layers work like this:

Graph Layer What It Contains How AI Uses It What Brands Must Do
Entity Graph Structured, low-ambiguity representations of who/what you are Confirms brand identity before including in answers Canonical entity home, consistent structured data, Wikipedia/Wikidata presence
Document Graph Indexed content with AI-extractable authority signals Sources specific claims and statistics Citation architecture — content engineered for AI extraction with quotable standalone facts
Concept Graph LLM associations between topics, categories, and brands Determines which brands are "top of mind" for a given concept Consistent vocabulary use, category ownership signals, earned media in Tier 1 sources

Most brands are optimizing for the document graph at best — publishing content they hope AI will extract. On that framework, the entities AI chooses most reliably are the ones operating across all three at once: a clearly defined entity, extraction-ready content, and concept associations reinforced through earned media from sources AI engines weight heavily.

This is the core of what AuthorityTech calls the MR Stack — the layered architecture that earns AI citations systematically rather than accidentally. Entity optimization is Layer 2 of the Stack, and it's the one most brands are still skipping entirely.

Why Concentration Is Accelerating

The 92% increase in concentration among the experts Authoritas tracked is a measurement, not an explanation. Search Engine Land's reading is that experts who actively manage their digital footprint are pulling away. Three dynamics plausibly compound each other:

Training data feedback loops. When AI systems are fine-tuned or updated, they use citation data from previous versions as signal. Entities with high citation frequency in early AI outputs carry that authority signal forward. The advantage isn't linear — it compounds with each model update. GraphRAG research published February 2026 confirms that deterministic inference paths through entity-enriched graphs produce significantly more consistent citation patterns than document-only retrieval.

Earned media amplification. Muck Rack's three What Is AI Reading? editions show that 82-89% of cited links came from earned media, the sources brands neither own nor pay for, not brand-owned content. Brands with Tier 1 media placements — Forbes, TechCrunch, WSJ, Bloomberg — are getting those placements recognized as authoritative corroboration for their entity signals. Brands without earned media have weak entity corroboration, regardless of how much they've published on their own domain.

Structured data scarcity. The percentage of websites with proper schema markup, consistent entity metadata, and AI-readable structured content is remarkably low. As AI systems increasingly reward structured signals, the gap between optimized and unoptimized entities widens rapidly. The 293% HHI jump in the tracked expert pool shows winner-take-most movement; scarcity of properly structured entities is one plausible contributor, not something that study measured. Search Engine Journal's 2026 enterprise SEO report identifies entity consistency as the single highest-leverage optimization gap in AI search.

The Citation Window: How Long Does It Stay Open?

The window analogy isn't alarmism — it's structural. Here's the reality: ChatGPT has about nine hundred million weekly users. Previsible tracked AI-referred sessions across 19 GA4 properties and found them up 527%, from 17,076 to 107,100, in the first five months of 2025 (its write-up describes the comparison both as January to May and as against the same months of 2024). Usage is scaling quickly, and the citation share of a well-optimized entity is still measurable and claimable today.

Eighteen months from now, it won't be. The brands and experts who've built entity authority through 2026 will be so deeply embedded in AI knowledge graphs — and so consistently reinforced through their citation feedback loops — that displacing them will require the kind of sustained earned media investment most brands can't commit to. The window isn't closing slowly. It is narrowing as AI usage grows.

What does "claiming entity authority before the window closes" actually mean in practice? It means five specific actions, executed in sequence.

5 Actions to Claim Entity Authority Now

1. Establish Your Canonical Entity Home

AI systems need a single authoritative source to resolve your brand's identity. This is your "entity home" — typically your main website, but specifically a page structured so AI can extract who you are, what you do, who you've helped, and why you're authoritative. That page needs Organization schema markup with complete properties: name, description, url, logo, foundingDate, founders, sameAs links to social profiles and Wikipedia. Inconsistency across these properties creates entity ambiguity that results in citation exclusion. Athena HQ's GEO analysis identifies entity home establishment as the foundational step before any other GEO investment — without it, all subsequent optimization produces diminished returns.

2. Build Cross-Platform Entity Corroboration

AI knowledge graphs look for corroboration — multiple independent sources confirming the same entity signals. Wikipedia, Wikidata, Google Business Profile, LinkedIn Company Page, Crunchbase, and authoritative industry directories all serve as corroboration nodes. Each one that reflects consistent entity information strengthens your resolution confidence across AI systems. Missing or inconsistent corroboration is why brands that dominate Google still disappear in AI answers. AWS's Bedrock AgentCore documentation explicitly describes how enterprise AI agents traverse entity graphs using `sameAs` relationships and canonical identifiers — the same mechanism consumer AI search uses for brand resolution.

3. Create Citation Architecture in Your Content

The Citation Architecture approach means engineering content specifically for AI extraction — not just for human reading. This requires quotable standalone sentences (10-20 words with specific data points), Key Takeaways sections with extractable statistics, FAQ structures that map to AI question-answering patterns, and structured comparison tables that AI engines love to reproduce. Generic long-form content doesn't get cited. Content structured for extraction does.

4. Invest in Tier 1 Earned Media as Entity Corroboration

A Forbes feature, TechCrunch mention, or Bloomberg profile is more than a traditional PR win. In the context of AI entity resolution, it's corroboration from a source that AI systems treat as authoritative. When an AI engine encounters your entity in a query context, it checks for corroboration in sources it trusts. Brands with consistent Tier 1 earned media presence pass that check. Brands without it fail it — and get skipped in favor of entities that pass. Across Muck Rack's three What Is AI Reading? editions, 82-89% of cited links came from earned media, the sources brands neither own nor pay for: the independent corroboration AI systems draw on for entity resolution (more in the Machine Relations research).

5. Audit and Close Your Citation Gap

Your Citation Gap is the delta between how you rank on Google and how often you're cited in AI answers. You can measure it by querying AI platforms directly for your category keywords and tracking how often your brand appears vs. competitors. Conductor's Share of Model framework provides a structured methodology for measuring AI citation frequency across platforms. AuthorityTech's AI Visibility Audit quantifies this gap and identifies which graph layers are underoptimized. Brands that audit their gap first can prioritize entity investments where the leverage is highest — rather than guessing.

What the Concentration Data Means for Your Category

The 59.5% concentration figure comes from one pool of 143 digital-marketing experts, not a market-wide average, and Authoritas did not measure other categories. Read it as an early signal of how fast a top tier can pull away in a field where AI engines already answer who-to-trust questions, not as a number for your category. Search Engine Land's February 2026 authority analysis notes that in recent Google core updates, larger, well-known brands have consistently outperformed smaller sites that were technically strong but lacked brand authority.

But here's the counterintuitive truth: experts and brands outside the top 10 aren't necessarily doomed. Concentration at the top doesn't mean the remaining 40.5% of that pool's citability is worthless — it means the competition for the remaining share is happening among those paying attention. The citation market isn't binary (cited vs. not cited). It's a share-of-voice model. February 2026 analysis from KEYT/Stacker makes the same case in its headline: AI didn't kill SEO — it killed average content. Specific, citation-ready content is what is left to compete on.

At AuthorityTech, we've built the infrastructure to help brands move from citation invisibility to citation authority — combining 8 years of Tier 1 media relationships with entity optimization built for the Machine Relations era. The window is still open. Barchart's 2026 GEO agency analysis identifies entity clarity and structured data as the two dimensions separating agencies delivering measurable AI visibility gains from those still running traditional SEO playbooks.

How GEO, AEO, and SEO fit within Machine Relations

These disciplines aren't competing alternatives — they represent different layers of the same system. Machine Relations is the full architecture that contains each of them:

DisciplineOptimizes forSuccess conditionScope
SEORanking algorithmsTop 10 position on SERPTechnical + content
GEOGenerative AI enginesCited in AI-generated answersContent formatting + distribution
AEOAnswer boxes / featured snippetsSelected as the direct answerStructured content
Digital PRHuman journalists/editorsMedia placementOutreach + storytelling
Machine RelationsAI-mediated discovery systemsResolved and cited across AI enginesFull system: authority → entity → citation → distribution → measurement

GEO and AEO are tactics within Layer 4 (Distribution) of the Machine Relations stack. They matter — but they operate on top of a foundation they cannot build on their own.

Frequently Asked Questions

What is entity optimization in AI search?

Entity optimization is the process of structuring your brand's digital identity so AI search engines can clearly resolve, verify, and cite you. Unlike traditional SEO which focuses on keywords and backlinks, entity optimization establishes who you are through structured data, consistent cross-platform signals, and earned media corroboration — the three elements AI knowledge graphs require before including a brand in generated answers.

Why did expert citability concentrate so rapidly?

The measurement is clear; the cause is not. In the Authoritas pool, the top 10 experts' share of weighted citability rose from 30.9% to 59.5% between December 2025 and February 2026, and the HHI rose 293%. One plausible mechanism is a feedback loop in which established entities get cited more, which reinforces their authority, which gets them cited again.

How is entity optimization different from GEO (Generative Engine Optimization)?

GEO — a distribution tactic within Layer 4 of the Machine Relations framework — focuses on optimizing individual pieces of content for AI extraction — structured formats, quotable data points, FAQ sections. Entity optimization works at the brand identity layer — establishing who you are across AI knowledge graphs so that content, when found, gets attributed to a clearly resolved entity. Both are necessary: GEO without entity optimization produces content that gets cited but not credited. Entity optimization without GEO produces a well-defined brand that lacks extractable content. The MR Stack integrates both as complementary layers.

What is the Citation Gap?

The Citation Gap is the delta between a brand's traditional search visibility and its AI citation frequency. A brand ranking #1 on Google for a keyword may appear in 0% of AI answers for that same query, while a lower-ranked competitor with stronger entity signals may appear in a large share of them. The Citation Gap identifies this disconnect and quantifies where investment in Machine Relations will drive the highest return on AI visibility.

How long does the entity optimization window stay open?

The window is narrowing as AI usage grows — Previsible's sample of 19 GA4 properties reported AI-referred sessions up 527% in the first five months of 2025. The only concentration measurement here is the Authoritas expert pool; no category-level figure exists in this data. The practical answer: citation share that earlier movers accumulate can compound in their favor, so starting entity work sooner is likely to be cheaper than starting later.

The Bottom Line

The entity concentration crisis isn't a future warning. It's a present measurement. Among the 143 digital-marketing experts Authoritas tracked, the HHI jumped 293% in two months and the top 10 now hold 59.5% of weighted citability. The knowledge graph infrastructure behind AI answers, a market forecast to reach $4.46 billion by 2030, is being built right now on entity signals that brands either have or don't have. The U.S. Chamber of Commerce's 2026 agentic AI report says, in a retail-focused piece, that a brand which isn't making detailed information about its products visible to AI search channels won't show up in the AI recommendations driving sales.

The brands that understand this — that entity authority is the new domain authority, that AI citations are the new first-page rankings, that the machine gatekeeper era has already begun — are building moats that will compound for years. The brands that don't understand it are watching their share of the most important distribution channel in the history of their industry get locked up by competitors who moved first.

Machine Relations isn't a trend you get ahead of. It's a market structure you either participate in — deliberately and systematically — or get excluded from by default.

The window is open. Build your entity. Close your Citation Gap. The clock is running.

For a concrete assessment of where your brand stands in your category's citation market, start with the AI Visibility Audit. You'll see exactly where you are in the concentration curve — and what it will take to move up it.

Related reading: Citation Economy: How Earned Media Shapes AI Visibility | How to Track ChatGPT, Perplexity and Gemini Traffic in GA4