Primary Research for Hedge Funds: How to Use Expert Networks, Surveys, and Channel Checks to Build and Challenge Investment Theses

Every fundamental hedge fund analyst knows the feeling: you've read the 10-K, built the model, digested the sell-side notes, and you still don't have enough conviction to size the position. The financials tell you what happened. Primary research tells you what's happening — and what's about to happen.

This guide is a practical operating manual for hedge fund analysts who want to get more out of their primary research process. We'll cover the full toolkit — expert calls, B2B surveys, channel checks, and transcript libraries — and show you how to structure each around a specific investment thesis. We'll also address the parts most guides skip: how to synthesise across sources, when to use which method, and how to build a repeatable research operating model that doesn't burn 50 hours of analyst time per name.

What Is Primary Research for Hedge Funds?

Primary research is the process of gathering first-hand, qualitative and quantitative insights — through expert interviews, B2B surveys, and channel checks — to build, validate, or challenge investment theses on public or private companies.

It's important to be precise about what we mean, because "primary research" gets conflated with "expert network calls" far too often. Expert networks are one delivery mechanism. The full toolkit includes:

• Expert calls (consultations): One-on-one conversations with industry practitioners, former executives, customers, or technical specialists — typically 30–60 minutes.
• B2B surveys: Structured quantitative surveys of industry participants (procurement managers, end-users, channel partners) designed to collect statistically meaningful data on a specific thesis question.
• Channel checks: Systematic outreach to customers, suppliers, distributors, or competitors of a target company to verify demand signals, pricing trends, or market share dynamics.
• Transcript libraries: Pre-recorded and transcribed expert interviews available on-demand, searchable by company, sector, or topic.
• Done-for-you research: End-to-end research delivery where a provider handles the entire process — from scoping and expert sourcing to synthesis — and delivers finished, actionable outputs.

The best analysts don't rely on any single method. They combine them deliberately, using each where it adds the most value.

Why Primary Research Matters More Now Than Ever

If you're a fundamental analyst in 2025, the case for rigorous primary research has never been stronger — and the cost of doing it poorly has never been higher. Here's why:

The hedge fund industry is at peak momentum

Global hedge fund AUM climbed to an all-time high of $4.74 trillion by mid-2025, fuelled by the largest quarterly inflows since 2014. The industry recorded back-to-back years of double-digit returns for the first time since the 2009–10 post-financial crisis rebound — with discretionary equity strategies delivering 17.1% returns and 5.7% alpha in 2025. Long/Short Equity accounted for 21% of fund searches, making it the top strategy. This is the segment most reliant on primary research.

Sell-side research is declining

Sell-side coverage has become increasingly scarce and less influential. The research you could once rely on from brokers is thinner, more consensus-driven, and often delayed. Institutional investors have turned to expert networks and proprietary primary research to fill the gap.

Everyone has the data — the edge is in synthesis

In 2026, you can't get an information edge just by having the data. Alternative data feeds, satellite imagery, credit card panels — they're available to anyone willing to pay. The edge now comes from synthesising faster and asking better questions. Primary research is one of the few remaining sources of genuinely differentiated insight, because the quality of the output depends on the quality of the input: your thesis, your questions, your ability to triangulate.

AI is automating the easy parts

By late 2025, 95% of hedge fund managers allowed employees to use AI tools. AI is automating scheduling, note-taking, and summarisation. But it cannot design the right questions, interpret the nuance of an expert's hesitation, or weigh one data point against a contradictory signal from a different source. The parts of primary research that require human judgment are becoming more valuable, not less.

Widening valuation dispersion favours active managers

Elevated market volatility from policy shifts, geopolitical tensions, and economic uncertainty is creating a more favourable environment for stock pickers. Wider dispersion means more opportunities for alpha — but only if you have the conviction to take differentiated positions. That conviction comes from primary research.

The Primary Research Toolkit: Methods and When to Use Each

Not every research question demands the same tool. Here's how to think about the five core methods:

Decision framework: which tool, when?

Use expert calls when you need depth on a narrow question, when the answer depends on operational experience that can't be surveyed (e.g., "What would it take for Company X to displace the incumbent in this account?"), or when you need to probe and follow up in real time.

Use surveys when you need quantitative backing for a qualitative hypothesis — especially for pricing studies, NPS comparisons, competitive perception mapping, or adoption rate estimation. Surveys are the most underutilised primary research tool among hedge funds. Expert calls get all the attention, but custom surveys provide statistically defensible data that PMs often find more convincing.

Use channel checks when you need to verify real-time demand signals: Are customers actually buying? Is pricing holding? Are competitors gaining share? Channel checks are particularly powerful for earnings preview work and short thesis development.

Start with transcript libraries when you're early in your research and want to map the landscape before committing to live calls. They're useful for identifying the right questions to ask, not for answering them definitively.

Use done-for-you research when you're covering multiple names, running parallel workstreams, or simply can't afford the 30–50 hours of analyst time that a self-directed expert call programme demands. Done-for-you providers handle the entire process — scoping, expert sourcing, interviews, synthesis — and return a finished deliverable.

How to Structure Primary Research Around an Investment Thesis

This is the section that matters most. The difference between high-value primary research and wasted expert calls comes down to one thing: structure. Here's the framework.

Step 1: Start with the thesis, not the topic

The most common mistake analysts make is approaching primary research with a vague brief: "I want to learn about the competitive landscape in industrial automation." That's not a research question. That's a semester of coursework.

Top hedge fund analysts approach primary research with specific hypotheses tied to their investment thesis. They articulate what they believe, why, and what would need to be true (or false) for the thesis to work.

Example: "We believe Company X is gaining share in the mid-market ERP segment because their product is 40% cheaper than the incumbent and customers report faster implementation times. We need to validate whether this pricing advantage is sustainable and whether implementation quality holds at scale."

That's a thesis. It can be tested. It can be proven wrong. It tells a research provider (or your expert network) exactly what kind of expert you need and exactly what questions to ask.

Step 2: Map thesis to specific questions

Break your thesis into 3–5 core assumptions. For each assumption, write 2–3 specific questions that would confirm or disconfirm it. These questions become the backbone of your discussion guide and survey instrument.

For the ERP example above:

• Assumption 1: Company X's pricing advantage is sustainable → "What are the key cost drivers behind Company X's pricing? Could incumbents match this pricing if they chose to? Have you seen any pricing changes in the last 6 months?"
• Assumption 2: Implementation quality holds at scale → "What has been your experience with Company X's implementation? How does it compare to [Incumbent]? Have you heard of implementation failures or delays?"
• Assumption 3: Customer switching costs are low enough to enable share gains → "What would it cost your organisation to switch from [Incumbent] to Company X? What's holding back faster adoption?"

Step 3: Select the right sources and methods

Now match your questions to the right primary research method and expert profile:

• Questions about operational experience → Expert calls with former employees, current customers, channel partners
• Questions about market-wide pricing or adoption → B2B survey of procurement decision-makers
• Questions about real-time demand → Channel checks with distributors and resellers

Be specific about the expert profile you need. "Someone who knows the ERP market" is too broad. "A VP of IT at a mid-market manufacturing company who evaluated Company X in the last 12 months" is actionable.

Step 4: Execute with discipline

Whether you're running calls yourself or briefing a provider, maintain rigour:

• Write a proper discussion guide — 10–15 structured questions mapped to your assumptions
• Share context with the expert (within compliance boundaries) so they can prepare
• Take structured notes mapped to your thesis assumptions, not just a stream-of-consciousness transcript
• After each call, score the expert's knowledge (relevance, recency, depth) so you can weight their input appropriately

Step 5: Synthesise across sources

This is where most analysts fall down. You've done 10 expert calls, run a 200-person survey, and completed 15 channel checks. Now what?

Build a synthesis matrix:

• Rows = your thesis assumptions
• Columns = each data source (expert calls, survey results, channel checks, public data)
• Cells = what each source says about each assumption (confirming, disconfirming, ambiguous)

This forces you to see where sources agree, where they conflict, and where you still have gaps. It protects against recency bias (over-weighting the last call you had) and anchoring on the most confident-sounding expert.

Step 6: Update conviction or kill the idea

Primary research should move you in one of three directions:

1. Higher conviction: Evidence supports the thesis across multiple sources → size the position, refine the model
2. Lower conviction: Evidence is mixed or contradictory → reduce position size, identify what additional data would resolve the uncertainty
3. Kill: Evidence clearly disconfirms a critical assumption → move on. This is one of the highest-value outcomes of primary research. Killing a bad idea early saves far more money than confirming a good one.

Running Effective Expert Calls

Expert calls remain the backbone of most hedge fund primary research programmes. Here's how to make each one count.

Design your discussion guide like it matters — because it does

The discussion guide is the single highest-leverage input in an expert call. Top analysts write 10–15 structured questions that map to specific assumptions in their model. They share the guide with the expert network to improve matching — because the better the network understands what you're really asking, the better the expert they'll find.

Structure your guide in three sections:

1. Background and context (2–3 minutes): Confirm the expert's role, tenure, and relevance. Don't skip this — it tells you how much weight to give their answers.
2. Core thesis questions (20–30 minutes): The structured questions mapped to your assumptions. Start broad, then narrow. Leave room for follow-up.
3. Open-ended exploration (5–10 minutes): "What am I not asking that I should be?" This is where unexpected insights surface.

Avoid the most common mistakes

• Winging it: Going into a call without a discussion guide produces unfocused, unrepeatable conversations.
• Leading the witness: "Don't you think Company X is losing share?" will get you the answer you want, not the answer you need.
• Treating all experts equally: An expert who left the industry three years ago should carry less weight than one who's actively in-market. Score and weight accordingly.
• Running too many calls without synthesis: 15 unstructured calls create noise, not signal. Five well-targeted calls with rigorous synthesis will outperform them every time.

Synthesising across multiple calls

After each call, log your findings against your thesis assumptions using a consistent format. After completing a batch of calls, review the aggregate:

• Where do experts agree? (Potential consensus view — is it already priced in?)
• Where do they disagree? (This is often where the real insight lives — understand why they disagree.)
• What surprised you? (Surprises often signal information asymmetry worth exploiting.)

Surveys and Channel Checks: The Underused Tools

B2B Surveys

Surveys are the most underutilised weapon in the hedge fund analyst's arsenal. While expert calls give you narrative depth, surveys give you statistical weight. PMs find it easier to act on "72% of procurement managers said they're considering switching from Incumbent to Company X" than "the three experts I spoke to thought Company X was gaining traction."

When to use surveys:

• Pricing perception and willingness-to-pay studies
• Net Promoter Score comparisons across competitors
• Feature adoption and product satisfaction rankings
• Purchase intent and switching probability
• Market sizing validation (bottom-up, from actual buyers)

Design principles:

• Keep surveys focused — 15–20 questions maximum
• Use a mix of closed-ended (quantitative) and open-ended (qualitative) questions
• Define your sample carefully: who are you surveying, and does their perspective actually test your thesis?
• Target a minimum of 50 respondents for directional data, 100+ for statistical significance

Channel Checks

Channel checks are particularly powerful for earnings preview work, short thesis development, and real-time demand verification. They involve systematic outreach to customers, distributors, resellers, and competitors of a target company.

When to use channel checks:

• Verifying whether a company's reported demand trends match reality on the ground
• Tracking pricing changes in real time
• Assessing inventory levels and sell-through rates
• Gathering evidence for short theses (customer dissatisfaction, competitive displacement)
• Pre-earnings intelligence gathering

Execution tips:

• Cast a wide net — talk to multiple points in the value chain (distributor, reseller, end customer)
• Use consistent questions across contacts so you can compare responses
• Focus on observable facts ("How many units did you sell last quarter?") rather than opinions ("Do you think Company X is doing well?")

Evaluating and Choosing Research Providers

Most investment firms maintain relationships with three to four expert networks to ensure comprehensive access and competitive pricing. But maintaining those relationships comes with a cost that goes beyond the invoice: the cost of your time managing them.

What to evaluate

When assessing providers — whether traditional expert networks, transcript libraries, or done-for-you research firms — evaluate on total workflow cost, not just the per-call or per-credit rate:

• Expert quality and match rate: How often does the first batch of expert profiles actually match what you asked for? What's the hit rate on calls that deliver actionable insight?
• Speed: How quickly can they arrange a relevant call? For hedge funds, speed matters — market events and earnings don't wait.
• Coverage depth: Can they source experts in niche sectors, emerging markets, and mid/small-cap companies? This is where most networks struggle and where the information asymmetry is greatest.
• Compliance infrastructure: Expert networks are evaluated on their ability to prevent material non-public information exchange, manage conflicts of interest, and maintain audit trails. This is non-negotiable.
• Total analyst time required: Map out the full workflow for a typical project with each provider. How many touchpoints require your team's time? Can any steps be fully offloaded? Is the provider delivering raw access or finished work product?
• Output format: Do they give you raw transcripts, or synthesised insights? The former still requires hours of analyst time to turn into something actionable.

The three models compared

Self-directed expert networks (GLG, AlphaSights, Third Bridge, Guidepoint): You submit a request, review profiles, schedule calls, conduct the calls, and synthesise findings yourself. Maximum control, maximum time investment. For a typical workstream with 15–20 expert calls, expect 30–50 hours of analyst time.

Transcript libraries (Third Bridge Forum, AlphaSense/Tegus, Stream, In Practise): Search and read pre-existing transcripts on-demand. Low time investment, but you're limited to questions someone else asked. Useful for background research; insufficient for thesis-specific questions.

Done-for-you research providers (e.g., Woozle Research): You brief the provider on your thesis and questions. They handle expert sourcing, interviews, surveys, and synthesis — delivering finished, actionable research outputs. You maintain analytical control without the production burden. This model is gaining traction as analyst time becomes the scarcest resource.

Pricing models

Transaction-based models account for roughly 60% of engagements in the expert network industry, with subscription-based models making up the remaining 40%. Most networks use a credit-based system where you purchase credits upfront and draw them down per consultation. Done-for-you providers typically charge project-based fees for end-to-end delivery.

Compliance and MNPI: What You Need to Know

Compliance isn't a section to skim. The FrontPoint insider trading case demonstrated what happens when analysts and experts bypass proper channels — experts and analysts communicated outside the network, leading to material non-public information exchange and enforcement action.

Practical compliance principles

• Never ask for MNPI: Material non-public information is any information that could move a stock price and isn't publicly available. If an expert starts sharing it, stop the call.
• Use the mosaic theory correctly: You can combine individually non-material pieces of public and non-public information to form an investment thesis. But each individual piece must be non-material.
• Maintain audit trails: Document every expert interaction — who, when, what was discussed, what compliance disclosures were made. Your compliance team will thank you.
• Evaluate your providers' compliance infrastructure: Sophisticated compliance frameworks — including expert vetting, conflict-of-interest screening, call monitoring, and audit capabilities — are table stakes. If your provider can't demonstrate these, find one that can.
• Be especially careful with current employees: Experts who are currently employed at a target company or its direct competitors carry higher MNPI risk. Ensure your provider has robust screening processes for these engagements.

How AI Is Changing Primary Research Workflows

AI is transforming research workflows — but not in the way most vendor pitches suggest. Here's what's real and what's hype.

What AI does well today

• Transcript summarisation: AI can digest 50 expert transcripts on a ticker in minutes, identifying key themes, sentiment shifts, and recurring data points.
• Expert sourcing: Over 30% of expert networks now offer AI-enabled sourcing, improving match speed and quality.
• Pattern detection: The best tools help analysts spot anomalies and changes in tone, theme, or sentiment across large volumes of information — identifying what's changing, where pressure is building, and whether evidence supports or challenges the current thesis.
• Scheduling and logistics: Automating the low-value coordination work that eats analyst time.

What AI can't do (yet)

• Design the right questions: The quality of primary research depends on the quality of the hypothesis. AI can't formulate the thesis that determines which questions to ask.
• Interpret nuance: An expert's hesitation, the way they qualify an answer, the topic they conspicuously avoid — these carry signal that current AI models miss.
• Weight conflicting evidence: When two credible experts disagree, the judgment call about which to trust (and why) remains fundamentally human.
• Make investment decisions: Synthesis is not summarisation. Turning research into a conviction-level view on a position requires experience, judgment, and accountability that AI doesn't provide.

The risk to watch

AI creates a false sense of confidence. Summarising 50 transcripts feels productive. But if the underlying questions were poorly designed, or the expert sample was biased, the summary is just a well-formatted aggregation of noise. Garbage in, polished garbage out.

Building a Primary Research Operating Model

If you're an analyst covering 10–20 names, you can't run deep primary research on all of them. You need a system for deciding where to invest your research time and how to execute efficiently.

Prioritise ruthlessly

Allocate primary research effort based on two dimensions:

1. Conviction gap: How much uncertainty exists around your key thesis assumptions? If the answer is "a lot" and the assumptions are critical to the thesis, primary research is high-value.
2. Information asymmetry potential: How much differentiated insight could primary research yield? Mid-cap and small-cap names with thin sell-side coverage often offer more asymmetry than mega-cap names where consensus is already well-informed.

Focus your deepest primary research on your highest-conviction, highest-uncertainty positions. Use lighter-touch methods (transcript library scans, quick channel checks) for maintenance coverage on other names.

Build a repeatable workflow

1. Thesis documentation: Before any primary research begins, write down your thesis, key assumptions, and what would change your mind. One page, maximum.
2. Question mapping: Map each assumption to specific questions and the best method to answer them.
3. Provider selection: Match the workstream to the right provider and method. Not every project needs live expert calls — some are better served by surveys, channel checks, or done-for-you research.
4. Execution tracking: Track calls completed, survey responses received, and channel checks done against your original plan. Don't let projects drift.
5. Synthesis and scoring: After each research batch, update your synthesis matrix and re-score conviction. Document what you learned and what remains unresolved.
6. Post-mortem: Track hit rates and attribution over time. Did your primary research lead to better outcomes? Which methods and providers delivered the most signal? Measure it quantitatively.

Know when to offload

Self-directed expert calls are the default for most hedge fund analysts, but they don't scale. If you're covering 15 names and need primary research on three of them simultaneously, the maths breaks down fast. A single workstream with 15–20 expert calls costs 30–50 hours of analyst time — time you're not spending on your other positions.

This is where done-for-you research providers earn their fee. You maintain control of the analytical process — the thesis, the key questions, the investment decision — while offloading the research production process. The output is a synthesised deliverable, not a pile of transcripts you need to read over the weekend.

Putting It All Together

Primary research is not a cost centre. It's a conviction engine. When done well, it's the difference between a hunch and a high-conviction position — and between a well-timed exit and a slow bleed.

The analysts who generate the most alpha from primary research share a few traits: they start with a testable thesis, they match the right method to the right question, they synthesise across sources with discipline, and they measure their process over time. They also know when to do the work themselves and when to offload the production to focus on what they're actually paid for — making investment decisions.

The expert network market is now a $3 billion industry growing at 12% annually, with over 120 active providers worldwide. The options have never been broader. But more options don't automatically mean better research. The bottleneck isn't access — it's workflow. The funds that win are the ones that build a systematic primary research operating model, not the ones that make the most expert calls.

If you're looking to get more out of your primary research — or want to explore what a done-for-you model could look like for your team — get in touch with Woozle Research. We handle the entire research process end-to-end, so you can focus on the analysis, not the logistics.