Every fundamental equity analyst has access to the same 10-Ks, the same earnings transcripts, and increasingly, the same AI-generated summaries of both. In 2026, public information is not a competitive advantage. It's table stakes.
Primary research — first-hand data collection through expert interviews, B2B surveys, and channel checks — remains one of the few reliable ways to generate differentiated insight on public companies. It's how the best analysts build conviction on a long thesis, pressure-test a short, or identify an inflection point before the rest of the market catches on.
This isn't a new concept. But the environment around it has shifted meaningfully:
- AI is commoditising secondary research at speed. Roughly 86% of hedge fund managers now use generative AI tools in investment or risk workflows. When everyone has the same AI-powered summarisation of the same public data, the edge moves upstream — to the quality of the questions you ask and the proprietary data you collect.
- Transcript libraries are becoming a crowded trade. If every fund reads the same Tegus transcript, the informational advantage evaporates. Original, thesis-specific primary research is more differentiated than ever.
- Fee compression demands sharper research ROI. Management fees are dropping toward 1.1%. Every dollar spent on primary research needs to demonstrably contribute to alpha. You can't afford to waste a week on fifteen generic expert calls that tell you what you could have read in a sell-side note.
- Performance dispersion rewards the prepared. Discretionary equity generated 17.1% returns and 5.7% alpha in 2025, but dispersion between top and bottom-quartile managers was considerable. The gap between good and great is often the quality of the primary research underpinning each position.
This guide covers the full workflow: from scoping a research question tied to a specific ticker, through designing and executing a primary research program, to synthesising findings into actionable investment insight. Whether you run the research yourself via an expert network or brief an external team to do it for you, the methodology is what matters.
The Primary Research Toolkit: Expert Calls, Surveys & Channel Checks
Primary research for hedge fund analysts falls into three core methods. Each has distinct strengths, and the best research programs usually combine more than one.
Expert Calls (One-on-One Interviews)
Structured conversations with industry practitioners — former executives, customers, channel partners, competitors, or technical specialists — who have direct knowledge relevant to your investment thesis.
When to use them:
- You need qualitative, nuanced insight that doesn't exist in public filings (e.g., "Why did Company X lose that contract?")
- You're exploring a new sector or business model and need to build a mental map quickly
- You want to validate or challenge a specific assumption with someone who has first-hand experience
Strengths: Depth, ability to follow threads in real-time, high signal when the expert is well-matched.
Limitations: Time-intensive (scheduling, conducting, synthesising), subject to expert quality variance, single data points can mislead without triangulation.
B2B Surveys
Quantitative data collection from a structured sample of industry participants — typically customers, users, IT decision-makers, or channel partners — designed to measure sentiment, adoption, spending intent, or competitive positioning at scale.
When to use them:
- You need a quantitative signal (e.g., "What percentage of mid-market CIOs are planning to increase spend on Company X's product?")
- You want to measure trends over time with repeatable methodology
- You need statistical validity to support a position in a high-conviction trade
Strengths: Scalable, quantifiable, repeatable, less subject to single-source bias.
Limitations: Requires careful design to avoid leading questions, respondent quality must be verified, less depth than interviews on any single response.
Channel Checks
The practice of collecting information from a variety of sources across a company's supply chain or distribution channels — distributors, retailers, resellers, implementation partners, field sales reps — to gauge real-time business momentum.
When to use them:
- You want to assess sell-through trends, inventory levels, or pricing dynamics ahead of earnings
- You're evaluating competitive shifts that won't show up in reported numbers for another quarter
- You need an early warning signal on acceleration or deceleration in a business
Strengths: Real-time, hard to replicate from public data, applicable across sectors (not just consumer/retail — software, healthcare, and industrials all benefit from channel work).
Limitations: Requires knowing who to talk to in the channel, compliance protocols must be rigorous, anecdotal data needs triangulation.
A critical misconception: channel checks are not just for consumer and retail analysts. Two-thirds of hedge funds with greater than $1.5 billion in assets perform channel checks, with over 45% using third-party channel research firms. If you're covering software, healthcare, or industrials and not running channel checks, you're leaving edge on the table.
How Primary Research Complements Secondary and Alternative Data
Primary research doesn't replace desk research — it sharpens it. The best analysts use filings, transcripts, and alternative data (satellite imagery, credit card data, web traffic) to formulate hypotheses, then deploy primary research to test and refine those hypotheses with real-world signal. The sequence matters: secondary data tells you what is happening; primary research tells you why it's happening and what comes next.
Designing a Thesis-Driven Research Program
This is where most analysts get it wrong. They jump straight to scheduling expert calls without a clear research design. The result: ten calls that yield generic industry commentary and no real conviction change.
The best primary research programs start with the thesis, not the call.
Step 1: Define the Investment Question
Before you engage a single expert, write down the 3-5 specific questions that would change your conviction level on the position. These should be falsifiable and directly tied to the investment thesis.
Weak: "Tell me about the competitive landscape in enterprise security software."
Strong: "Is Company X losing share in the mid-market segment to Company Y, and if so, is the driver pricing, product gaps, or channel execution?"
The goal is to identify inflection points — positive or negative — before they are widely understood by the market. If your research question can be answered by reading a sell-side report, it's not specific enough.
Step 2: Scope the Right Respondent Mix
The respondent mix is the single biggest determinant of research quality. You need diverse perspectives that triangulate on the same question from different angles:
- Former employees of the target company (product, sales, engineering — not just C-suite)
- Customers (current and, critically, churned customers)
- Channel partners (resellers, VARs, system integrators, distributors)
- Competitors (sales reps and product managers at rival firms)
- Industry specialists (consultants, analysts, or operators with cross-company visibility)
Talking exclusively to former executives gives you one angle. Triangulating across customers, channel partners, and competitors gives you a three-dimensional picture of what's actually happening in the market.
A compliance note: the risk of an employee breaching a duty of confidentiality is lower if they are surveyed about other companies — competitors, suppliers, distributors — or about other firms' products, rather than their own employer's non-public information.
Step 3: Write an Effective Discussion Guide
A discussion guide is the structured set of questions you'll use across all interviews on a given research program. It serves two purposes:
- Consistency: Asking the same core questions across experts enables pattern recognition and triangulation.
- Efficiency: Keeps conversations thesis-relevant and prevents them from drifting into generic industry commentary.
A good discussion guide:
- Opens with context-setting questions (role, tenure, exposure to the topic)
- Moves quickly to thesis-specific questions
- Includes both open-ended probes and specific data points you're trying to validate
- Leaves room for unexpected signals (the best insights often come from follow-up questions)
- Ends with a "what are we missing?" prompt
Step 4: Determine Sample Size and Timeline
There's no universal rule, but as a practical guide:
- For a focused thesis question: 8-15 expert interviews across the respondent mix, or a survey of 50-150 qualified respondents, is typically sufficient to reach signal convergence (the point where additional interviews confirm rather than add new insight).
- For a channel check: 15-25 channel contacts across geographies or segments, depending on the distribution model.
- Timeline: A well-designed research program should be executable in 1-2 weeks. If it's taking longer, the scope is too broad or the sourcing is too slow.
Remember: more calls do not equal better research. Five tightly scoped, high-quality conversations with well-matched experts will outperform twenty generic calls every time.
Running the Research: Self-Service vs. Done-for-You
Once you've designed your research program, you have a fundamental choice: do you execute it yourself, or do you brief a team to do it for you?
The Self-Service Model: Expert Networks
Expert networks — GLG, AlphaSights, Third Bridge, Guidepoint, Tegus, and others — are structured platforms that connect investment professionals with industry experts. The expert network market hit approximately $3 billion in 2025, growing around 12% annually, with about 11,200 firms using these services.
The model works like this: you submit a request describing the type of expert you need, the network matches you with candidates from their database, compliance reviews the match, and you schedule and conduct the call yourself.
Key players and what differentiates them:
- GLG (Gerson Lehrman Group): Over 1 million experts across 19 global offices, revenue exceeding $400 million. The largest by revenue, with the broadest expert base.
- AlphaSights: More than 500,000 experts, nine offices worldwide, revenue over $300 million. Strong in consulting and private equity engagements.
- Third Bridge: Claims the largest expert count at 1.5 million, with eight offices. Offers Third Bridge Forum, a proprietary transcript and interview library, alongside live calls.
- Guidepoint: Exceeds 1 million experts through 14 offices across three continents. Differentiates through flexible engagement options and competitive pricing.
- Tegus: Combines expert calls with a library of pre-recorded interviews. Built specifically for investment professionals — hedge funds, PE, and asset managers.
- Capvision: Focused on Asia-Pacific coverage with 360,000 experts. A go-to for APAC-specific research needs.
How to evaluate an expert network:
- Expert quality and relevance: The biggest value driver is how fast they can get you to the right expert without compliance risk. A provider that is fast but off-target is unusable.
- Coverage breadth: Does the network have depth in your sector and geography? If you cover APAC healthcare, a network with no Asian experts is not useful.
- Compliance infrastructure: Pre-call compliance screening, documentation protocols, cooling-off periods for former employees.
- Response time: How quickly can they produce a qualified match? In an earnings-driven workflow, 48 hours may be too long.
- Cost structure: Transaction-based (per-call billing, often $500-$5,000+ per hour for the expert) vs. subscription models. Transaction-based models account for approximately 60% of expert interactions in the hedge fund and PE sectors.
Most institutions work with multiple expert network providers to diversify access and reduce dependency on any single platform.
The Done-for-You Model
Here's the question most analysts don't ask themselves often enough: should I be doing this work at all, or should I brief a team that does this every day?
A fundamental long/short analyst typically covers 10-30 names. Running 10-15 expert calls per thesis — scheduling, conducting, taking notes, synthesising — consumes entire weeks. Multiply that across an active portfolio, and you're spending the majority of your time on research execution rather than research thinking.
The done-for-you model flips this: you brief a research team on what you need to know, and they handle the entire process end-to-end — research design, respondent sourcing, expert interviews, synthesis, and deliverable. No call scheduling, no transcript review, no synthesis. The output is a finished research product, not a pile of transcripts.
When done-for-you research makes more sense:
- Time-constrained theses: You need answers before next week's earnings, not next month.
- Breadth of coverage: You're initiating on a new name and need 15+ perspectives quickly — doing all those calls yourself isn't realistic given your other coverage responsibilities.
- Synthesis quality: The hardest part of primary research isn't getting on a call — it's synthesising 15+ disparate expert perspectives into a coherent, thesis-relevant conclusion. A team that does this daily will produce a tighter deliverable.
- Differentiation: If you're reading the same Tegus transcripts as every other fund, your research isn't differentiated. Custom, thesis-specific research programs produce proprietary insight.
This is what we do at Woozle Research. Investment teams brief us on the thesis, and we go and do the primary research for them — expert interviews, B2B surveys, channel checks — delivering finished, actionable research outputs. Not expert access. Not transcripts. Research. If that's a model that could save you time while sharpening your edge, we should talk.
Compliance and the Mosaic Theory
Compliance isn't optional in primary research — it's existential. The SEC has sharpened its focus on how investment advisers manage MNPI, and enforcement is active. In January 2025, the SEC charged a hedge fund manager with MNPI failures related to a consultant engagement. General counsels across the industry have become far more cautious about keeping their teams away from the line.
The MNPI Framework
Material Non-Public Information (MNPI) is confidential information that has not been disclosed to the public and could significantly influence an investor's decision to buy, sell, or hold securities. Examples include upcoming earnings results, planned acquisitions, major management changes, or significant contract wins or losses.
Trading on MNPI is illegal. Full stop.
The Mosaic Theory
The mosaic theory allows analysts to legally piece together information from multiple public sources and non-material, non-public sources to form investment conclusions. The key requirement: no material non-public information is used in the analysis.
A critical misconception: the mosaic theory does not permit a firm to trade while any person at that firm possesses MNPI about the issuer in question. Assembling a mosaic of non-material pieces is legal; adding one material piece to the mosaic is not.
Practical Compliance Protocols
- Pre-call compliance review: Before every expert interaction, expert details should be reviewed by compliance to screen for conflicts (e.g., the expert is a current employee at a covered company, subject to a cooling-off period, or bound by active NDAs).
- Opening disclaimer: Every call should begin with an explicit statement that you do not want to receive any MNPI, trade secrets, or information subject to confidentiality obligations.
- Real-time documentation: Detailed notes from every call, documenting what was discussed and what sources of information were referenced.
- Post-call compliance check: After each interview, the analyst should confirm to compliance that no MNPI was received.
- Cooling-off periods: Former employees of covered companies should be subject to cooling-off periods (typically 6-12 months post-departure) before being engaged as experts on their former employer.
Common Compliance Traps
- The "should have known" risk: Most analysts rely on the expert to refuse to speak if they have a confidentiality obligation. But a company might argue the analyst "should have known" their employees had a duty of confidentiality.
- Undocumented calls: If you can't prove what was discussed, you can't prove it wasn't MNPI.
- Informal channels: Research conversations that happen outside of formal compliance-reviewed channels (a casual dinner, a LinkedIn message) carry the same legal risk as a formal expert call, but without the documentation.
Here's the counterintuitive insight: compliance is a competitive advantage, not just a cost centre. Funds with robust compliance infrastructure can be bolder in their primary research programs because they have the guardrails to operate confidently. The funds that pull back from primary research out of compliance anxiety are leaving alpha on the table.
Synthesising Research into Investment Insight
This is the bottleneck. Not getting on a call — synthesising 15+ disparate expert perspectives into a coherent, thesis-relevant conclusion that actually changes your conviction level on a position.
Moving from Raw Notes to Thesis-Relevant Conclusions
After completing your interviews, you should be able to answer:
- What is the consensus view among my sources? (And does it differ from the market's consensus?)
- Where is there disagreement, and why? (Disagreement is where alpha lives — understand what's driving the divergence.)
- What data points directly address my original 3-5 thesis questions?
- What did I learn that I didn't expect? (Unexpected findings are often the highest-value output of a primary research program.)
- Has my conviction changed, and in which direction?
Triangulation
Useful investment insight rarely comes from a single source. The real value comes from identifying where multiple, independent sources converge on the same signal — or where they diverge in a way that reveals something the market hasn't priced.
A former VP of Sales at Company X tells you they're losing deals in the mid-market. A channel partner confirms win rates are declining. A competitor's sales rep says they're picking up accounts that used to be Company X's. A customer survey shows declining satisfaction scores. That's convergence. That's signal.
One disgruntled former employee telling you the product is terrible? That's an anecdote. Without triangulation, it's noise.
What a "Good" Research Deliverable Looks Like
Whether you're synthesising your own research or evaluating a deliverable from an external provider, the output should include:
- Executive summary: 1-2 paragraphs that state the key finding and its investment implication.
- Thesis scorecard: How each of your original research questions was answered, with supporting evidence.
- Key themes: 3-5 major themes that emerged from the research, with representative quotes or data points.
- Dissenting views: Where sources disagreed, and what explains the disagreement.
- Data appendix: Survey results, respondent profiles, or detailed interview summaries for those who want to dig deeper.
If your research output is a stack of raw transcripts, you haven't done primary research — you've done data collection. The research is in the synthesis.
Sector Playbooks
Software / SaaS
The public software market's shift to SaaS has made traditional point-in-time financial metrics less useful. Metrics like ARR, NRR, and LTV have become central to valuation, yet quarterly financials are lagging indicators. By the time a slowdown in net revenue retention becomes public knowledge, the market has often already reacted.
What to research: Customer adoption and expansion trends, competitive win/loss dynamics, renewal and churn patterns, pricing changes, implementation partner sentiment.
Who to talk to: IT decision-makers and end users at customer accounts, VAR and SI partners, former sales and customer success leaders, competitive sales reps.
Key questions: Are customers expanding usage or consolidating vendors? How is the product positioned vs. alternatives in new deals? Is the sales motion shifting (e.g., from enterprise to mid-market, or vice versa)?
Healthcare / Pharma
What to research: Prescribing trends, formulary positioning, clinical outcomes data (from practitioners), competitive pipeline impact, patient access and reimbursement dynamics.
Who to talk to: Physicians (prescribers), pharmacy benefit managers, payer executives, KOLs (key opinion leaders), medical affairs professionals at competitor firms.
Key questions: Is adoption of Drug X accelerating or plateauing? How are payers positioning it on formulary relative to alternatives? What's the real-world efficacy experience vs. clinical trial data?
Industrials / Consumer
What to research: Distributor and retailer sentiment, inventory levels, pricing and promotional trends, demand signals from end customers, supply chain dynamics.
Who to talk to: Distributors, retailers, field sales reps, supply chain managers, purchasing managers at customer accounts.
Key questions: Are orders accelerating or decelerating vs. last quarter? Is the company gaining or losing shelf space? Are channel partners increasing or decreasing inventory?
The AI Layer: How Technology Is Changing Primary Research Workflows
AI adoption across hedge funds has been dramatic: 95% of hedge fund managers allow employees to use multiple AI applications as of late 2025. Funds leveraging generative AI are achieving 3-5% higher annualised returns compared to non-adopters, with the most significant benefits in equity hedge strategies.
But here's the nuance: AI is accelerating research workflows, not replacing primary research itself.
Where AI Adds Real Value
- Transcript search and pattern recognition: Tools like AlphaSense and Third Bridge AI can surface relevant passages across thousands of earnings transcripts, broker research reports, and expert call transcripts in seconds. This is genuinely useful for hypothesis generation — identifying threads to pull before designing a primary research program.
- Auto-summarisation of secondary research: Generating first-draft summaries of filings, transcripts, and news. This frees analyst time for higher-order thinking.
- Expert sourcing and matching: Over 30% of expert networks now offer AI-enabled expert sourcing. Faster matching means faster research execution.
- Survey design and analysis: AI can assist with survey instrument design and rapid analysis of quantitative results.
Where AI Falls Short
- Research design: Deciding what to ask, who to ask, and how to interpret ambiguous findings still requires human judgment — specifically, the judgment of someone who understands both the investment thesis and the industry dynamics.
- Expert quality assessment: AI can match keywords, but determining whether an expert actually has the specific, current knowledge you need requires contextual human evaluation.
- Nuanced synthesis: GenAI models are only as good as their input data. If the inputs are noisy, incomplete, or biased, the outputs can be misleading. One hedge fund attempting to train a custom LLM on internal analyst notes found that inconsistent formatting and jargon-laden commentary led to unreliable summaries.
- Compliance judgment: Determining whether a piece of information constitutes MNPI requires human legal and contextual judgment that AI cannot reliably provide.
The bottom line: AI helps you move faster through the research funnel, but it doesn't replace the design, execution, or judgment layers of primary research. The funds gaining real edge from AI are using it to compress the cycle — not to skip the work.
Measuring ROI on Primary Research Spend
With management fees compressing and research budgets under scrutiny, you need to be able to justify every dollar spent on primary research. Here's how to think about it:
Linking Research to Portfolio Decisions
Track, at minimum:
- Conviction change: Did the research program increase, decrease, or change the direction of your conviction on the thesis? If it didn't change anything, the research was either poorly designed or you were asking questions you already knew the answers to.
- Position action: Did the research lead to initiating, sizing up, trimming, or exiting a position?
- Timing value: Did the insight arrive early enough to act on before the market repriced?
Evaluating Vendor Performance
Whether you're using expert networks, transcript libraries, or done-for-you research providers, track:
- Match quality: What percentage of expert matches were actually relevant and useful?
- Speed: How quickly did the provider deliver? For hedge fund timelines, days matter.
- Insight density: How many actionable insights per engagement? Five calls that yield one useful data point is a different ROI than five calls that each move the needle.
- Compliance incidents: Any near-misses or issues should be tracked and factored into provider evaluation.
Where to Go From Here
Primary research remains the sharpest tool in the fundamental analyst's toolkit — but only when it's done with rigor. The analysts generating the most differentiated insight in 2026 are the ones who:
- Start every research program with a falsifiable thesis question
- Design the respondent mix before picking up the phone
- Use structured discussion guides for consistency and triangulation
- Maintain airtight compliance protocols at every step
- Invest their time in research design and interpretation — and consider outsourcing the execution when speed and quality demand it
The expert network market is more competitive than ever, AI tools are accelerating every stage of the workflow, and the bar for what constitutes "differentiated research" keeps rising. The analysts who thrive will be the ones who treat primary research as a disciplined process — not a series of phone calls.
If you're looking to sharpen your primary research process — or hand off the execution entirely so you can focus on what to do with the insight — get in touch with the Woozle Research team. We run expert interviews, B2B surveys, and channel checks for hedge fund analysts every day, and we'd welcome the chance to show you what a finished research deliverable looks like.
