We analyzed what AI actually believes about buyers across 25 industries. The findings should change how you think about content strategy.
Most brands create content based on assumptions. Internal workshops produce three to five buyer personas that feel right, and those personas drive the content calendar for the next year. Maybe longer.
The problem is that AI systems - the ones increasingly mediating buyer research - operate on a completely different model of who your buyers are and what they want. And that model is built from patterns across millions of data points, not from a two-hour stakeholder session.
We wanted to know: what does AI actually believe about buyers? Not what marketers assume. Not what CRMs suggest. What emerges when you synthesize and structure buyer behavior data at scale?
So we built the largest buyer persona dataset ever published. And the patterns it reveals should make every content strategist reconsider their funnel.
What We Built and Why
PersonaGen-593K is a dataset of 593,181 AI-generated buyer personas spanning 339 industries (25 primary verticals). Each persona includes structured behavioral data: who they are, what they search for, what questions they have, what goals drive them, and what frustrates them.
The raw numbers:
| Attribute | Volume |
|---|---|
| Total personas | 593,181 |
| Industries covered | 339 (25 primary verticals) |
| Search queries extracted | ~3 million |
| Information need statements | ~3 million |
| Goal statements | ~1.8 million |
| Pain point statements | ~1.8 million |
We started with approximately 40 million raw persona descriptions from four public datasets (NVIDIA, BSC-LT, Orange, and Tencent). Through GPU-accelerated deduplication and semantic filtering, we reduced that to about one million unique descriptions. Then we enriched each one with structured behavioral attributes using a large language model, ultimately producing 593,181 complete, validated personas.
This is not a survey of 500 people. It is a structured synthesis of what AI models understand about buyer behavior at a scale that would take traditional research methods years and millions of dollars to approximate.
The Finding That Should Rewrite Your Content Strategy
78.7% of buyer searches are informational. Only 4.3% are transactional.
Read that again. Across nearly 3 million search queries in the dataset, nearly four out of five are questions like "How do I evaluate X?" or "What are the best approaches to Y?" - not "Buy X" or "Subscribe to Y."
The remaining 17% are commercial - comparison queries like "Which vendor should I choose?" That is the consideration phase. Important, but still not transactional.
What This Means in Practice
| Search Intent | Share of Queries | What Buyers Are Doing |
|---|---|---|
| Informational | 78.7% | Learning, researching, understanding options |
| Commercial | 17.0% | Comparing vendors, evaluating alternatives |
| Transactional | 4.3% | Ready to buy, subscribe, or sign up |
Most content strategies are built around the bottom of the funnel. Product pages. Pricing comparisons. Feature matrices. Demo requests. That is the 4.3%.
The 78.7% - the questions buyers are actually asking - often goes underserved. And that is precisely the content AI systems pull from when forming recommendations.
If your brand does not have strong informational content that answers the questions buyers ask before they are ready to buy, AI will recommend someone who does.
Five Industry Patterns Worth Knowing
The dataset reveals distinct behavioral signatures across industries. Here are the patterns most relevant to content and marketing leaders.
1. EdTech Dominates in Volume - and Opportunity
EdTech accounts for 24.3% of all personas in the dataset (36,152). That is not because we oversampled it. It emerged naturally from the source data, reflecting the sheer volume and diversity of learning-oriented buyer behavior.
EdTech personas concentrate around certification pathways, online learning flexibility, and skill development. If you operate in education technology, adjacent learning services, or professional development, this is the largest addressable persona pool in the dataset. The content opportunity is enormous - and so is the competition for AI mindshare.
2. Consulting Has the Highest Commercial Intent
While the overall dataset skews informational, Consulting personas buck the trend. They show the highest commercial search intent at 28% - nearly double the dataset average of 17%.
Consulting buyers know what they need. They are comparing options, not exploring the concept. Their pain points center on client acquisition and differentiation - which means content that helps them evaluate options (benchmarks, case studies, ROI frameworks) will outperform general thought leadership.
3. Healthcare Faces a Unique Compliance Challenge
Healthcare personas exhibit the highest concentration of compliance-related information needs at 38%. They search for credentialing, regulatory requirements, and evidence-based practices.
For brands in healthcare or health-adjacent industries, this carries a specific risk: AI systems generating recommendations about healthcare topics face a higher bar for accuracy. Inconsistent or poorly structured health content becomes a liability vector when AI surfaces it in professional contexts.
4. Manufacturing Wants Process, Not Vision
Manufacturing personas cluster around operational efficiency (29% of goals) and struggle most with tool complexity (24% of pain points). Their searches focus on process optimization and automation.
This audience does not respond to aspirational messaging. They want specifics: implementation guides, integration documentation, comparison matrices with technical specifications. Content that skips the vision and goes straight to the operational "how" wins in this vertical.
5. Work-Life Balance Is the Universal Pain Point
Across all 25 industry verticals, work-life balance emerged as the dominant pain point with 107,000 mentions. It appears as a top-five concern in 72% of industry segments.
This is a cross-industry truth that most B2B content ignores entirely. Brands that acknowledge the human dimension of professional decisions - time constraints, cognitive load, competing priorities - speak to the reality their buyers live in, not just the category they buy in.
The Demographics AI Assumes
The dataset also reveals what demographic profile AI associates with different industries and roles:
- Gender distribution: 52.2% female, 45.4% male, 2.3% non-binary. Near-balanced, but not uniform across industries.
- Age pattern: Bimodal, with peaks at 18-25 (27.9%) and 55-75+ (41.4% combined). The mid-career cohort that most B2B marketing targets (25-55) accounts for just 30.8%.
- Market context: 71.5% B2C, 22.5% B2B, 4.4% B2B2C, 1.5% B2G.
The age distribution is particularly notable. If your content strategy assumes a 35-to-50-year-old decision maker as the default buyer, AI models may associate your industry with a very different demographic profile. That mismatch between your assumed audience and AI's assumed audience affects which content gets surfaced and to whom.
What This Means for Your Brand
If you are in EdTech or professional development
You are operating in the most crowded persona space. Differentiation requires specificity - not "we help people learn" but "we help mid-career UX designers transition to instructional design with employer-recognized credentials." The personas show that certification pathways and flexible scheduling are the dominant search triggers. Generic learning content will not cut through.
If you are in Consulting or professional services
Your buyers have the highest commercial intent in the dataset. They are comparison-shopping, not browsing. Invest in content that supports evaluation: transparent pricing frameworks, methodology comparisons, client outcome data with specifics. The "thought leadership first" playbook that works in other verticals may underperform here because your buyers are already past the awareness stage.
If you are in Manufacturing or industrial
Skip the vision. Lead with operational detail. The personas show that tool complexity is the primary frustration and process optimization is the primary goal. Content that demonstrates measurable efficiency gains with clear implementation steps will match how AI characterizes manufacturing buyer intent.
If you are in Healthcare
Structure your content for accuracy and compliance. AI systems are especially cautious with health-related recommendations, and the persona data confirms that your buyers prioritize credentialing and regulatory information. Unstructured or ambiguous content is more likely to be deprioritized by AI in this vertical than in any other.
Where This Research Comes From
PersonaGen-593K was developed by researchers at the Estonian Entrepreneurship University of Applied Sciences (EUAS) and Rankfor.AI, published as a peer-reviewed paper in Springer's Discover Artificial Intelligence journal.
The methodology involved aggregating four public persona datasets (~40 million raw descriptions), applying GPU-accelerated deduplication to remove 89.5% of duplicates, then using semantic embedding models to eliminate a further 76.2% of near-duplicates. The surviving one million unique descriptions were enriched with structured behavioral attributes and validated through statistical analysis.
The full dataset, methodology, and analysis code are publicly available for independent verification.
The Practical Takeaway
Buyer behavior data at this scale tells a consistent story: most buyers are asking questions, not looking to purchase. The brands that win AI recommendations are the ones whose content answers those questions clearly, specifically, and in the language buyers actually use.
If 78.7% of your potential buyers are in learning mode, and your content strategy allocates 80% of resources to product-focused bottom-funnel material, there is a structural mismatch between what you produce and what AI recommends.
Closing that gap is not a content volume problem. It is an alignment problem. And the first step is understanding what AI believes your buyers want.
Check how AI perceives your brand's buyer personas. Use the Persona Matcher to compare your assumed personas against what AI actually recommends for your industry. Or run your brand through the AI Visibility Monitor to see where you stand.
Based on "PersonaGen-593K: A Large-Scale Dataset of AI-Generated Buyer Personas for Consumer Information-Seeking Behavior Research" by Dmitrij Zatuchin (EUAS) and Daniil Dzemesjuk (Rankfor.AI), preprint, submitted to Springer Discover Artificial Intelligence, 2026.