What Expert Networks Actually Do

An expert network sells one thing: a conversation with somebody who already knows the answer.

That is the entire product. The platforms, the compliance stacks, the databases with a million profiles in them, all of it exists to make that one conversation happen faster, more legally, and with a better-matched person on the other end of the line than the buyer could have managed alone.

The industry is roughly thirty years old, worth somewhere between two and five billion dollars depending on who is counting, and remains poorly understood by most of the people writing checks to it. This is a guide to what it actually is, what it gets used for, and how to tell whether you are getting the thing you paid for.

Inex One, the platform many institutional buyers use to manage expert network spend, puts the market at approximately three billion dollars in 2025, growing around twelve percent annually since 2023 (Inex One). IBISWorld sizes the United States industry alone at 1.8 billion dollars in 2026 (IBISWorld). Business Research Insights values the global market at 4.86 billion dollars for the same year (Business Research Insights). Growth estimates across published research run from roughly seven percent to sixteen percent.

Those are not small discrepancies. A market cannot plausibly be both 1.8 billion and 4.86 billion in the same year.

The gap is definitional rather than a measurement error. "Expert network" once meant a single service: arrange a paid phone call between an investor and an operator. It now describes firms that sell survey panels, transcript libraries, artificial-intelligence-led interviews, data labeling, and full research projects delivered as a report. Count the transcript libraries and the survey panels and the number roughly doubles. Count only the calls and it does not.

When a network quotes a price per interaction, the first question is which of those businesses you are buying from, because the economics and the failure modes are completely different. A transcript library is a content product with near-zero marginal cost. A custom-recruited call is a labor product where somebody has to go find a person who does not yet know they are for sale.

Gerson Lehrman Group launched in 1998 on a straightforward arbitrage. Investors needed to understand industries they did not work in. People who worked in those industries were willing to talk for money. Nobody had built the connective tissue, the compliance wrapper, or the billing mechanism to make that a repeatable transaction.

Two structural facts followed from that origin and still shape the market.

First, the original customer was the hedge fund, and the original use case was public-markets research. That is why the compliance architecture looks the way it does, and it is why the industry's worst moment arrived through that channel.

Second, the model is not capital intensive. There is no factory. A competent operator with relationships can start a network, which is why there are now roughly 150 firms operating globally (Silverlight Research) and why the leader's share has eroded steadily. Inex One's data shows Gerson Lehrman Group falling from roughly 51 percent of the market to roughly 24 percent over a decade. The market grew and the incumbent did not grow with it.

Today the largest firms by revenue are Gerson Lehrman Group, AlphaSights, Third Bridge, Guidepoint, and AlphaSense following its acquisition of Tegus. Below them sit a long tail of regional specialists, sector specialists, and boutiques.

Commercial due diligence. The single largest use case, and the one that drives most private equity spend. A sponsor is evaluating a target. The management presentation says the company has a defensible position, low churn, and a pricing advantage. Expert calls are how you find out whether that is true from people who are not being paid to sell you the asset: former executives at the target, current and former customers, competitors, distributors, and suppliers along the value chain. The question being answered is almost never "what does this industry do." It is "is the specific claim in this specific deck actually true."

Market entry and sizing. A corporate or a sponsor-backed platform wants to know whether a geography, a channel, or an adjacent category is worth entering. Published market research will give you a total addressable market figure. It will not tell you why the last three entrants failed, which is usually the more decision-relevant fact and is only available from people who watched it happen.

Competitive intelligence. Understanding a competitor's cost structure, roadmap, channel strategy, or internal dysfunction. This is where compliance discipline matters most, and where the line between market intelligence and material nonpublic information has to be actively policed rather than assumed.

Customer and channel checks. Frequently the highest-value calls in a diligence. Not "do customers like the product" but "what happens at renewal, who else is in the room, and what would make you switch." Buyers consistently underweight this angle relative to former-employee calls, and it is usually where the surprises live.

Technology and vendor evaluation. Whether a target's technical advantage is real or is workflow automation with a marketing layer on it. Operators who have personally evaluated, bought, or ripped out the relevant systems can settle in twenty minutes what a vendor demonstration cannot settle at all.

Post-acquisition value creation. Less visible than diligence and growing quickly. Once the deal closes, the sponsor needs operators who have run the specific playbook being contemplated: a pricing reset, a route-to-market change, a manufacturing footprint consolidation. This is advisory work delivered in one-hour increments.

Public markets research. The founding use case. Channel checks, supply chain verification, thesis testing ahead of an earnings print. It remains a meaningful share of industry revenue and carries the heaviest compliance burden by a wide margin.

Across all of these, the segmentation that matters commercially is not sector, it is spend concentration. Consulting and market research account for roughly half of industry spending. Capital markets, meaning asset managers, private equity, and hedge funds, account for roughly 42 percent. Corporates make up about 45 percent of clients by count but only around 10 percent of spend (Inex One). Most firms that use expert networks are corporates. Most of the money is not.

The mechanics are unglamorous and they are where quality is won or lost.

A brief arrives. It contains a thesis, a set of questions, and usually a list of target companies split into currents and formers. The network's first job is to decode it, because briefs are written by people who know what they want and not always what is findable. A brief asking for "current executives at the top five competitors" is asking for something that is either non-compliant, unavailable, or both.

That decoding produces angles. An angle is a specific population: former vice presidents of operations at named competitors, current procurement leaders at named customers, distributors in a named geography. A well-scoped project usually has three to six angles, and the angles are the actual deliverable of the scoping conversation.

Then sourcing. Two models exist and the difference is the main quality variable in the industry. Database matching runs the brief against an existing roster and returns the closest matches. It is fast and it is why buyers complain about seeing the same faces across multiple projects. Custom recruiting goes and finds people who are not currently in anybody's database, which is slower and produces a materially better fit for a specific question.

Screening comes next, and it is the step most often skipped. A profile that matches on title tells you almost nothing. Screening means putting the project's own questions to the expert before the client's time is spent, and reading the answers for whether this person was close enough to the problem, recently enough, to be useful. A network that passes profiles through on title alone is selling you a calendar invitation.

Then compliance clearance, scheduling, the call itself, and, on well-run engagements, a feedback loop where the client tells the network what worked so that subsequent angles sharpen.

Every expert network runs a compliance apparatus. Almost every buyer treats it as paperwork. It exists because of a specific set of events.

In February 2011 the Securities and Exchange Commission charged a group of consultants and employees connected to Primary Global Research, a California expert network, with insider trading. Technology company employees had been moonlighting as paid consultants without their employers' knowledge and passing material nonpublic information about companies including Advanced Micro Devices, Dell, and Marvell to hedge fund clients. Experts were paid between 150 and 1,000 dollars an hour. The associated trading generated more than 30 million dollars (SEC). Multiple participants received prison sentences.

That case, and the wider insider trading enforcement wave around it, produced the compliance architecture the industry now runs on: employer restrictions, blackout periods around earnings, prohibitions on discussing an expert's current employer, recorded attestations, and client-side policies governing who may speak to whom.

Most of the visible compliance apparatus controls for identity and for procedure. It verifies that the person on the call is who they claim to be and that both parties agreed not to break the law. Those are worth having. Neither one controls for whether the answer you receive is true, which is the failure mode that actually costs a client money. A perfectly compliant call with an expert who is three years stale, or who is reasoning from a model rather than from experience, will pass every check in the stack and still send a deal team in the wrong direction.

Pricing has two layers: what the expert is paid, and what the client is charged.

Expert compensation is set per hour and is driven by seniority, scarcity, and how often the person is asked. Operating-level and mid-management experts commonly sit in the low hundreds. Senior executives at large companies routinely command four figures. Genuinely scarce profiles, meaning a named role at a named company where perhaps a dozen people worldwide qualify, price higher still.

Clients pay a multiple of that, either per call or through a credit system. Credits are where the opacity lives: a credit is worth a different dollar amount at different networks, and the same nominal "one credit call" can represent materially different underlying economics. Ask what a credit costs in dollars and what happens to unused credits at renewal.

Two structural notes. First, most consulting-firm work runs pay-as-you-go, which means the network is absorbing sourcing cost with no protection on its time. That is why a disorganized project is expensive for the network and why disorganized clients get slower service, whether or not anybody says so. Second, larger buyers increasingly move to prepaid retainers drawn down over time, which smooths the relationship for both sides and usually improves the service level.

If the product is a phone call, why not just make the phone call yourself? LinkedIn exists. Your associates are capable.

For a single conversation, that objection holds, and buyers should act on it more often than they do. If you need to talk to one person and you know who they are, go talk to them.

It stops holding at volume and under a clock. A commercial diligence needing twenty conversations across six angles inside three weeks is a sourcing operation, not a networking exercise. Someone has to identify several hundred candidates, reach them cold, screen them against the thesis, clear them for compliance, negotiate rates, schedule across time zones, and replace the ones who cancel. That is a full-time job for several people and it is the actual thing being purchased. The call is the deliverable. The sourcing is the product.

The second, less comfortable answer is compliance. An investment firm whose analysts arrange their own paid conversations with industry insiders, with no attestation trail and no employer-restriction screening, has created a regulatory problem that is considerably more expensive than the network fee.

Four signals, in rough order of how much they predict.

Were you asked hard questions during scoping? A network that takes the brief without pushing back on it is not going to catch the angle you got wrong. The scoping call is where most of the value is created or lost.

Are you seeing profiles you have seen before? Recycled profiles across projects indicate database matching. That is not disqualifying, and for broad market questions it is often sufficient, but you should know which product you bought.

Does the screening go beyond title? Ask what the expert was asked before the profile reached you. If the answer is nothing, the fit assessment is being outsourced to you and you are paying for it in call slots.

What happens when they cannot find the person? The honest answer is sometimes that a profile does not exist in a reachable form. A network that says so early is more useful than one that spends your window producing near-misses.

Three developments will determine what this industry looks like in three years.

The first is in-house sourcing. Mid-sized consulting firms and some sponsors have begun building recruiting capability internally, moving into a lane the networks owned outright. Partly cost, partly a preference for not distributing a map of what they are looking at across four vendors.

The second is the artificial intelligence entrant wave. Considerable capital is going into networks promising to automate matching and, in some cases, the interview itself. The plausible outcome is that the sourcing and screening layer gets faster while the value migrates further toward the scarce conversation that automation cannot reach. Worth watching whether any of these firms is founded by someone who has run the delivery side, because the failure modes are almost all on that side.

The third is definitional. If transcript libraries and survey panels keep growing faster than call volume, "expert network" stops being a useful category, and buyers will need to price the three businesses separately. The disagreement about market size is the leading indicator.