How expert networks actually work — a buyer's explainer

If you've never run a project through an expert network, the pricing is opaque and the sourcing happens behind a curtain. Most of what's online is marketing copy.

This is a plain explainer for first-time buyers. It covers what an expert network is, how the workflow runs, how vendors price it, where quality actually comes from, and how compliance works — then answers the questions buyers ask most.

What an expert network is

An expert network is a matchmaking service: it connects buyers who have a question with people who've actually done the work. A consultant working on a commercial diligence wants 30 minutes with a former product manager at the target company. An investment team wants two hours with the head of supply for a category they're underwriting. The network finds the operator, vets them, schedules the call, handles compliance, and bills the buyer for the time.

The industry traces back to the late 1990s. Gerson Lehrman Group (GLG) was founded in 1998 — the first firm to organize the model as a productized service rather than an ad-hoc referral practice ¹. Today, the market includes large global incumbents (GLG, AlphaSights, Guidepoint, Third Bridge), tech-enabled players (Dialectica, proSapient, NewtonX), transcript-and-call hybrids (notably AlphaSense, which acquired Tegus in 2024), and a newer wave of AI-driven matching platforms.

The five-step workflow

Most expert engagements run through the same five steps, regardless of vendor.

Anatomy of an expert engagement

1. 1

   Brief intake

   0–30 min

   The buyer submits a research question and constraints. Better networks treat this as an artifact; lower-touch ones treat it as a search query.

2. 2

   Sourcing

   1–4 hr

   Candidates are pulled from the network's internal expert database, plus enrichment from data providers. Match quality depends on how much of the brief actually informs the search.

3. 3

   Screening

   Same day

   Candidates are pre-screened against the question — sometimes by a project manager, sometimes by an AI-driven workflow, sometimes by both. Notes become part of the shortlist.

4. 4

   Compliance

   Inline or pre-call

   Conflict checks, NDAs, restricted-topic prompts, and consent confirmation. Best-in-class networks run these inline with sourcing; others run them as a separate gate.

5. 5

   Engagement

   30–90 min

   The call (or survey, or written response) happens. Transcripts, summaries, and audit logs go back to the buyer.

How expert networks make money

Pricing models vary, but four dominate:

• Subscription with call credits. Buyer pays an annual fee for a fixed number of expert calls; overages are billed per call. Most common at large incumbents.
• Pay-as-you-go per call. Each expert engagement is billed separately, usually at an hourly rate that covers the network's cut plus the expert's compensation. Common at boutiques and newer platforms.
• Project-based. A single fee for a defined scope of work — for example, "20 calls plus a synthesized writeup." Common in tech-enabled and AI-driven networks.
• Marketplace fee. Some aggregators take a transaction cut on calls routed through their platform; the underlying network keeps the rest.

Pricing is the least transparent part of the industry. Contracts are bespoke, rate cards are not published, and per-call costs vary substantially with expert seniority, geography, and exclusivity. There's no honest single number to quote here — ask the vendor for a written rate card before committing, and benchmark across at least two suppliers.

Where quality comes from — and where it doesn't

The buyer-side question that matters most isn't "how big is the network?" — it's "what does the network actually do to find the right expert?" Three components drive quality:

• The expert database. Raw size matters less than freshness and segment depth. Bigger networks have more names. Better networks have more current, consent-verified names in the segment you care about.
• The search workflow. A search that reads the brief and decomposes it into perspectives, company universes, and screens will outperform a keyword-style search even on a smaller database.
• The screening layer. Whatever the matching technology, candidates have to be pre-screened against the specific question before they make the shortlist. Networks that voice-screen on substance produce different shortlists than networks that voice-screen for availability.

What does not drive quality (despite being marketed heavily):

• Network size as a standalone claim. "Largest network" is a procurement signal, not a quality signal — the candidate that matters is the one for your question.
• Speed alone. A same-day shortlist is worthless if the candidates miss the question. Speed is a multiplier, not a replacement.
• "AI-powered" without specifics. Almost every vendor's site says this now. Ask which step it touches: sourcing? screening? compliance? rationale generation? All four? The answer tells you whether AI is doing real work or branding.

Why compliance is part of the workflow, not a gate

Expert networks operate in a regulatory environment that has been actively enforced for over a decade. The defining case is the 2011 prosecution of Primary Global Research (PGR), a US expert network whose sales executive and several experts were convicted of insider trading for passing material non-public information (MNPI) to hedge fund clients. The case reshaped industry practice: today, every reputable network runs explicit MNPI prohibitions, restricted-topic prompts, consent and NDA workflows, and engagement audit trails ².

For buyers, the compliance layer matters in three concrete ways:

• MNPI prevention. The expert is contractually prohibited from sharing non-public material information about a current employer. The network's job is to make this prohibition operational — restricted-topic prompts, screening for company-specific MNPI risk, and audit logs.
• Conflict checks. The network screens candidates against the buyer's stated exclusions (current employer, recent employer, board affiliations, equity events) before the candidate appears on a shortlist.
• Audit trail. Every engagement should produce an exportable record of the brief, the candidates considered, the screening notes, the conflict checks, and the call itself. If the buyer's compliance team can't reconstruct the engagement from the network's records, the network is carrying risk on the buyer's behalf.

Common questions from first-time buyers

Is using an expert network legal?

Yes. Expert networks are a routine part of professional research workflows at consulting firms, investment funds, and corporate strategy teams. The legal exposure is not the network itself — it's specific behaviors during the engagement (passing MNPI, violating an expert's employer NDA, soliciting information the expert is contractually prohibited from sharing). Standard compliance controls — restricted-topic prompts, NDAs, consent forms — exist to keep engagements on the right side of those lines.

What's MNPI and why does it matter?

MNPI stands for "material non-public information" — information about a company that hasn't been publicly disclosed and would, if disclosed, influence an investor's decision. US securities law treats trading on MNPI as insider trading. Expert networks must ensure their experts do not share MNPI about current or recent employers; that is the single largest compliance risk in the workflow.

How fast should I expect a shortlist?

For a well-scoped brief at a major network, expect a first-pass shortlist next day. AI-driven platforms can return a shortlist in under an hour for many brief shapes. Speed is rarely the binding constraint, though. Match quality is. A fast shortlist of the wrong people loses to a slower shortlist of the right one.

What does a typical call cost?

Pricing varies dramatically by expert seniority, geography, and the buyer's subscription tier — and no vendor publishes a public rate card. The right answer is to ask each vendor for a written rate sheet during procurement and to benchmark across at least two suppliers before committing. Treat any single number you hear secondhand as a rumor.

Is the call covered by my NDA?

Standard practice: yes, but you should confirm. Reputable networks operate under an MSA that includes confidentiality terms covering the engagement, the brief, and the expert's identity. The expert separately signs a confidentiality agreement before the call. If either is missing, ask before booking.

Can I ask experts about their current employer?

About the company in general, often yes; about non-public matters (financials, strategy, customer pipelines, M&A), no. The network should brief the expert on which topics are off-limits and reinforce the boundary during the screening step.

When does a transcript library beat fresh calls?

Transcript libraries — notably AlphaSense's Tegus product, Third Bridge Forum, and Guidepoint Insights — excel when the question is historical, broad-market, or has already been asked recently by another buyer. Fresh calls win when the question is novel, time-sensitive, or requires a specific operator perspective the library doesn't contain. Most projects use both: scan the library to triage the obvious, run fresh calls on the gaps.

The 2026 landscape

Three structural shifts shape what buyers are evaluating right now:

• Transcript libraries have moved from a feature to a category. AlphaSense's acquisition of Tegus in 2024 was the signal: reusable expert content is now sold as a standalone product, separate from live calls.
• AI-driven matching has moved from marketing claim to real product surface. A handful of platforms now publish written match rationale, run inline conflict checks, and decompose briefs automatically — capabilities that didn't exist as productized workflows three years ago.
• Compliance language is converging. MNPI prohibitions, screening, and audit logs are now baseline everywhere. The differentiation is in how visible and exportable those artifacts are.

What to ask any vendor now: not "can you do compliance?" but "can you show me the audit log?" Not "do you use AI?" but "which step, and what artifact does it produce?"

Footnotes

1. GLG was founded by Mark Gerson and Thomas Lehrman in 1998. The company's own corporate history and contemporaneous coverage in the financial press describe it as the first firm to formalize on-demand access to subject-matter experts as a paid service. ↩

2. SEC Litigation Release No. 21806 (February 2011), United States v. Fleishman et al., S.D.N.Y. James Fleishman, a Primary Global Research sales executive, was convicted of conspiracy to commit securities fraud after coordinating MNPI tips from current-employee experts to hedge fund clients. The case is widely cited in compliance literature as the inflection point for expert-network compliance practice. ↩