AI Citation Analysis Spring '26 Update
Multi-dimensional analysis of AI citation patterns | Q4 2025 Data
Our new study evaluates 23 million+ AI citations across over 17 million unique URLs, this study reveals a critical insight for modern SEO: businesses have far more influence over AI-generated answers than previously assumed.
Nearly 91% of all AI citations originate from sources businesses can directly control, including owned websites, structured data, and authoritative brand properties. The findings highlight a major shift in AI search visibility, emphasizing that strategic content optimization, technical SEO, and data accuracy play a decisive role in how brands are referenced and surfaced in AI-powered search and answer engines.
You can read the full report here, or explore it interactivly below.
filter_list Filters & Dimensions
Start with a sector, then refine with question vectors, industries, and models.
Objective / Subjective
Branded / Unbranded
AI Models
insights Current View Insight
Loading citation data...
help Question Type Explorer
Click a question type to view detailed breakdown.
Objective Branded
Factual queries about specific brands
--
of citations
Objective Unbranded
Factual queries without brand mention
--
of citations
Subjective Branded
Opinion queries about specific brands
--
of citations
Subjective Unbranded
Opinion queries without brand mention
--
of citations
© 2026 Yext Research
layers Control Category Explorer
Click a category to highlight it across all visualizations.
Full Control
Owned Properties
--
of citations
Some Control
Managed Listings
--
of citations
Limited Control
User-Generated Content
--
of citations
No Control
Third-Party Content
--
of citations
© 2026 Yext Research
Overall Citation Controllability
Aggregate distribution of citation control levels
© 2026 Yext Research
Citation Controllability by Model
Control category distribution across AI models
© 2026 Yext Research
Full Control (Websites)
Some Control (Listings)
Limited Control (Reviews/Social)
No Control (News/Forums)
Sector Overview
Click any row to filter by sector
progress_activity Loading sector data...
© 2026 Yext Research
23M
Total Citations
17.2M
Distinct URLs
About This Data
Analysis of 23,014,316 citations across 17,192,798 distinct URLs reveals that businesses have significant control over how they appear in AI-generated answers. Nearly 91% of all citations come from sources businesses can directly influence.
Citation Volume by Model
| Model | Citations | Share |
|---|---|---|
| Perplexity | 8.76M | 38.0% |
| OpenAI (SearchGPT) | 6.58M | 28.6% |
| Gemini | 6.29M | 27.3% |
| Claude | 1.39M | 6.1% |
Citation Reuse Frequency
| Model | Avg. Frequency |
|---|---|
| Perplexity | 1.41x |
| Gemini | 1.40x |
| OpenAI | 1.26x |
| Claude | 1.09x |
How often each unique URL is cited on average
© 2026 Yext Research
menu_book
Guide
How to Read This Report
This report analyzes where AI models source their citations from when answering questions. Understanding these patterns helps marketers optimize their content strategy for AI visibility.
Understanding "Controllability"
"Controllable" citations are those from sources marketers can directly influence. This includes only:
| Control Level | Source Type | Marketer Action |
|---|---|---|
| Full Control | Owned websites & properties | Direct content updates |
| Some Control | Listings & directories | Claim & manage profiles |
Total Controllable = Full Control + Some Control
What's NOT Controllable
These sources are outside direct marketer influence:
| Control Level | Source Type | Why It's Difficult |
|---|---|---|
| Limited Control | Reviews & social media | User-generated content |
| No Control | News, forums, gov't sites | Third-party editorial |
While you can't control these directly, reputation management and PR can influence them indirectly.
Key Terms
| Citation | A URL referenced by an AI model in its response |
| Distinct URL | Unique web pages cited (same URL may be cited multiple times) |
| Reuse Frequency | How often the same URL is cited across responses (Citations ÷ Distinct URLs) |
| Objective Query | Factual questions: "What are the hours?", "Where is it located?" |
| Subjective Query | Opinion questions: "Is it worth it?", "What's the best?" |
| Branded Query | Questions mentioning a specific business name |
| Unbranded Query | Generic category questions without brand names |
AI Models Analyzed
| Model | Provider | Notes |
|---|---|---|
| Perplexity | Perplexity AI | Highest citation volume (38%) |
| SearchGPT | OpenAI | Prefers listing sources |
| Gemini | Highest controllability (93%) | |
| Claude | Anthropic | Most diverse sources |
© 2026 Yext Research
25.3%
Controllability Range
Controllability by Industry
Controllability ranges from 70.7% (Non-Profit) to 96.0% (Specialty Services). Service-based industries with strong local presence tend to have higher controllability.
Most Controllable Industries
| Industry | Controllable |
|---|---|
| Specialty Services | 96.0% |
| Automotive Retail | 95.9% |
| Dental & Vision | 95.0% |
| Logistics & Mailing | 94.3% |
| Specialty Retail | 93.5% |
Least Controllable Industries
| Industry | Controllable |
|---|---|
| Non-Profit | 70.7% |
| Industrial | 76.4% |
| Hypermarkets | 77.2% |
| Specialty Food & Cafes | 80.2% |
| Logistics (Finance) | 80.6% |
Top Sectors by Citation Volume
| Sector | Citations | Share |
|---|---|---|
| Retail | 5.50M | 23.9% |
| Healthcare | 4.62M | 20.1% |
| Business Services | 4.43M | 19.3% |
| Finance | 3.93M | 17.1% |
| Food & Beverage | 2.81M | 12.2% |
| Hospitality | 1.51M | 6.6% |
| Organizations | 0.16M | 0.7% |
Sector "No Control" Exposure
| Sector | Limited Control | No Control |
|---|---|---|
| Organizations | 3.2% | 14.9% |
| Hospitality | 7.0% | 7.5% |
| Food & Beverage | 11.1% | 4.2% |
| Healthcare | 3.8% | 4.2% |
| Retail | 3.3% | 3.4% |
Largest Sample Size by Volume
| Industry | Citations | Share |
|---|---|---|
| Automotive Retail | 2.90M | 12.6% |
| Specialized Physicians | 2.22M | 9.6% |
| Restaurants | 2.11M | 9.2% |
| Insurance | 1.98M | 8.6% |
| Maintenance & Repair | 1.60M | 7.0% |
Industries with Highest Review Exposure
| Industry | Review % |
|---|---|
| Specialty Food & Cafes | 12.3% |
| Restaurants | 10.7% |
| Maintenance & Repair | 10.0% |
| Construction | 9.5% |
| Professional Services | 7.1% |
Food service and home services industries have 2-3x the review exposure of average, making reputation management critical.
© 2026 Yext Research
90.6%
Controllable Citations
Controllability
Nearly 91% of all AI citations originate from either owned websites (Full Control) or managed listings (Some Control). This means businesses have significant opportunity to shape how they appear in AI-generated answers.
Controllability by Category (On Average)
| Control Level | Description | % of Citations |
|---|---|---|
| Full Control | Owned websites & properties | 46.9% |
| Some Control | Managed listings & directories | 43.7% |
| Limited Control | Reviews & social media | 5.5% |
| No Control | News, forums, government, other | 3.9% |
Controllability by Model (On Average)
| Model | Full Control | Some Control | Total |
|---|---|---|---|
| Gemini | 51.0% | 42.2% | 93.2% |
| OpenAI | 44.4% | 48.0% | 92.4% |
| Perplexity | 46.7% | 42.1% | 88.8% |
| Claude | 41.1% | 40.4% | 81.5% |
Model Source Preferences
| Model | Prefers | Websites | Listings | Reviews |
|---|---|---|---|---|
| Gemini | Websites | 51.0% | 42.2% | 3.5% |
| Perplexity | Websites | 46.7% | 42.1% | 7.0% |
| OpenAI | Listings | 44.4% | 48.0% | 3.4% |
| Claude | Balanced | 41.1% | 40.4% | 15.0% |
Key Insight: Gemini strongly prefers websites (+8.8pp), while OpenAI favors listings (+3.6pp). Claude pulls significantly more from reviews (15% vs 3-7% for others).
© 2026 Yext Research
56.7%
Objective
43.3%
Subjective
Question Types
Questions are categorized by intent (objective vs. subjective) and whether a brand is mentioned (branded vs. unbranded). Unbranded queries consistently show higher controllability.
All Question Types Ranked by Controllability
| Question Type | Example | Citations | Controllable |
|---|---|---|---|
| Objective Unbranded | "Best dentist near me" | 7.24M (31.5%) | 93.6% |
| Subjective Unbranded | "Is it worth getting braces?" | 6.02M (26.1%) | 91.4% |
| Objective Branded | "What are [Brand]'s hours?" | 5.80M (25.2%) | 89.4% |
| Subjective Branded | "Is [Brand] any good?" | 3.96M (17.2%) | 85.8% |
fact_check Objective Questions
Factual queries: hours, location, services, prices
| Share of Citations | 56.7% |
| Total Controllable | 91.7% |
| Full Control | 49.7% |
| Some Control | 42.0% |
| Limited + No Control | 8.3% |
reviews Subjective Questions
Opinion queries: reviews, recommendations, comparisons
| Share of Citations | 43.3% |
| Total Controllable | 89.2% |
| Full Control | 43.2% |
| Some Control | 46.0% |
| Limited + No Control | 10.9% |
Branded vs. Unbranded
| Metric | Unbranded | Branded |
|---|---|---|
| Total Citations | 13.26M (57.6%) | 9.76M (42.4%) |
| Full Control | 53.7% | 37.6% |
| Total Controllable | 92.6% | 87.9% |
Key Insight: Unbranded queries yield 16pp more Full Control citations. Invest in category-level content to capture these higher-controllability queries.
Why Subjective Has Lower Controllability
| Source Type | Objective | Subjective |
|---|---|---|
| Limited Control (Reviews) | 4.5% | 6.8% |
| No Control (News/Forums) | 3.8% | 4.1% |
Key Insight: Subjective queries pull 51% more review content. Reputation management is critical for businesses receiving opinion-based queries.
© 2026 Yext Research
