An expert network sells you an hour with someone who knows your market. A customer research platform gets you an hour with someone who buys in it. Those sound similar and they are not, and the difference decides whether a diligence process or a market-entry decision ends up grounded in evidence or in well-informed recollection.
Short answer: use an expert network when you need knowledge you cannot legally or practically obtain yourself, such as a competitor's internal economics, a supply chain you do not touch, a regulatory process you have never navigated, or a market where you have no customers yet. Use customer interviews when the question is what buyers actually value, why they switch, what they will pay, or why they churned, because in those cases the expert is a proxy and the customer is the source.
What expert networks are, and what they cost
Expert networks recruit and vet subject-matter experts, then broker paid one-to-one calls between them and clients, mostly investors, consultancies and corporate strategy teams.
The major players and the published estimates on them:
- GLG (Gerson Lehrman Group) is the largest in the world, with a network reported at over one million vetted experts. Third-party benchmarks put client rates at 1,500 to 2,000 US dollars per hour, while the lowest rate paid to some experts is around 250 US dollars.
- Guidepoint runs a database also reported at over one million experts across roughly 200 industries, with effective client rates benchmarked at approximately 1,200 US dollars per interview hour.
- AlphaSense took a different route. It acquired Tegus in July 2024 for 930 million US dollars, alongside a 650 million dollar funding round that valued the combined company at 4 billion US dollars. AlphaSense announced that ARR surpassed 400 million US dollars in March 2025, more than double the 200 million reported in April 2024, with over 6,000 customers including 88 percent of the S&P 100.
- Third Bridge and AlphaSights occupy the same broad space, selling a mix of live calls and a library of pre-recorded interview transcripts.
Two structural notes. First, none of these firms publish reliable rate cards, so every number above is an estimate from directories and buyer-side benchmarking rather than an official price. Second, the model is shifting from live calls toward content libraries: the Tegus acquisition added a library covering more than 35,000 public and private companies and over 150,000 transcripts of investor-led expert interviews, folded into an AlphaSense corpus of roughly 450 million searchable documents. AlphaSense seat pricing has been reported in the region of 10,000 to 20,000 US dollars per user per year, with expert call credits typically billed on top.
Independent estimates size the global expert network market at roughly 3 billion US dollars in 2025, growing at around 12 percent a year, with the United States accounting for about 55 percent of worldwide revenue at approximately 1.8 billion dollars. Treat these as directional; the category has no audited public reporting.
The compliance rule that removes the freshest information
This is the part of the model that buyers underweight, and it is not a scandal. It is the system working as designed.
Expert networks operate under a compliance framework shaped by the SEC's insider trading enforcement of the early 2010s, and MNPI, material non-public information, remains an active examination focus for the SEC's examinations division. In practice the framework rests on three pillars: real-time documentation of calls, MNPI surveillance, and cooling-off periods that bar recent insiders from being consulted about their former employer.
Now follow the consequence. The cooling-off period exists precisely because a person who left a company last month knows things that are material and non-public. So the network must exclude them. The expert you are permitted to speak to is, by construction, the one whose information has aged past the point of being dangerous.
That is the right rule. It is also a hard ceiling on freshness. You are buying recollection of a market as it was when the expert still had access to it, filtered through however many months or years the compliance regime requires and human memory adds on top.
Your own customers are under no such constraint. They can tell you what they are doing today, because it is their own behaviour and their own money.
The expert is not the customer
Strip away the pricing and the compliance mechanics, and one distinction remains.
An expert network sells testimony about a market. Customer interviews collect evidence from it.
Every layer of remove costs you something specific:
- Recall replaces observation. An expert describes what buyers wanted when they last sat across from them. A customer describes what they are choosing now, with the invoice in front of them.
- Narrative smoothing. Experts are paid to be articulate and they are good at it. Articulate answers are coherent, and coherence is exactly what you should be suspicious of, because real buying behaviour is messy, contradictory and full of constraints the expert has tidied away.
- Selection. The population of people who take paid consulting calls is not a random sample of people who know your market. It skews toward the between-roles, the recently departed and the professionally available.
- Nobody in the conversation can churn. An expert has no account with you, no renewal date and no budget line. They can be completely right about the category and still tell you nothing about whether your specific customers are about to leave.
None of that makes expert calls worthless. It makes them the wrong instrument for a large class of questions that buyers routinely point them at.
Where expert networks genuinely win
Be clear about this, because the category exists for good reasons.
- Markets where you have no customers yet. Entering an adjacent category, you have nobody to interview. An expert who ran that category for eight years is the fastest route to a mental model.
- Competitor internals. Nobody else can tell you how a rival's channel margin structure works or why their last product cycle slipped.
- Supply chain and channel checks. Distributors, procurement leads and manufacturing operators you have no relationship with.
- Regulatory and technical process knowledge. How an approval pathway actually behaves as opposed to how it is documented.
- Very small populations. When the total number of relevant buyers globally is 200, you are not running a research study, you are making phone calls, and a network can find the right ones fast.
- Investment diligence under time pressure. A deal team with two weeks needs named operators immediately, and the network's job is producing them.
For deal work specifically, the two approaches combine well: our playbook on commercial due diligence customer interviews and the guide to customer research for investors both assume you will triangulate operator testimony against actual customer evidence rather than choosing one.
Where customer interviews win
- What buyers actually value. Ask the people who chose, not the people who used to sell to them.
- Why deals were lost. Win-loss is a category where secondhand explanation is almost always wrong. The buyer knows why. See the win/loss analysis guide.
- Willingness to pay. Expert estimates of price sensitivity are opinion; buyer reactions to your actual price are data.
- Churn diagnosis. Only the churned account can explain the churn.
- Sizing grounded in demand rather than assertion. Top-down expert estimates are the weakest input in most market models. Our guide to TAM SAM SOM for product researchers covers building the bottom-up version.
The comparison
| GLG / Guidepoint / AlphaSights / Third Bridge | AlphaSense (post-Tegus) | Koji | |
|---|---|---|---|
| What you get | Live 1:1 calls with vetted experts | Searchable transcript library plus expert calls | AI-moderated interviews with your own customers |
| Who speaks | Former operators, category specialists | Previously recorded expert interviews | Actual current customers and buyers |
| Reported cost | Roughly 1,200 to 2,000 USD per hour | Roughly 10,000 to 20,000 USD per seat per year, calls extra | 29 or 79 euro per month; 3 credits per voice interview |
| Freshness | Constrained by cooling-off rules | Historical by definition | Today, in the participant's own words |
| Scale per study | A handful of calls | Large library, no new primary data | Dozens of parallel interviews |
| Analysis | Your notes | AI search over documents | Automatic thematic coding with traceable quotes |
| Best for | Inaccessible knowledge, competitor internals | Fast background and prior context | Demand-side evidence and diagnosis |
The arithmetic
Take a market-entry question. The conventional approach is 20 expert calls. At benchmarked rates of 1,200 to 2,000 US dollars per hour, that is roughly 24,000 to 40,000 US dollars and typically several weeks of scheduling, plus compliance chaperoning on each call.
Now price the demand-side counterpart. In Koji, a voice interview costs 3 credits and a text interview costs 1, with credits at 1 euro. Forty voice interviews with real buyers is 120 credits, they run in parallel rather than sequentially, and thematic coding is finished when fielding closes.
These are not substitutes and the comparison is not meant as one. The point is proportion. Most teams spend the entire research budget on the expensive proxy and nothing at all on the primary source, because expert calls feel like the serious option. Running both costs barely more than running one.
How to combine them properly
The sequence that works:
- Start with experts to build the map. Five to eight calls to learn the category structure, the vocabulary, the economics and who the players are.
- Turn what you learned into a question set. The expert calls tell you what to ask; they do not tell you what is true of your buyers.
- Go to the demand side at volume. Run 30 to 50 interviews with actual buyers. Use structured questions where you need comparability across respondents. Koji supports six types,
open_ended,scale,single_choice,multiple_choice,rankingandyes_no, so you get a ranked list of purchase drivers and the reasoning behind the ranking in the same study. The structured questions guide covers the mechanics. - Code and compare. Thematic analysis across the interviews shows you where buyers contradict the expert consensus. Those contradictions are usually the most valuable output of the whole exercise, because an expert consensus that every competitor also bought is not an edge.
- Go back to experts on the anomalies. Now the expensive hour is spent on a sharp question rather than an orientation session.
If you are running expert conversations yourself rather than through a network, our expert interviews guide covers planning, recruiting and running them without the brokerage layer.
Related reading: Koji vs CleverX compares B2B expert panel access directly, and Koji vs Kantar covers the agency side of the same buying decision. For the other two secondhand evidence sources, see what analyst reports actually measure and what software review sites can and cannot tell you.
Frequently Asked Questions
How much do expert networks cost per call?
None of the major networks publish rate cards, so all figures are third-party benchmarks. GLG client rates are commonly benchmarked at 1,500 to 2,000 US dollars per hour, and Guidepoint at approximately 1,200 US dollars per interview hour. Experts themselves receive a fraction of that, with the lowest reported expert rates around 250 US dollars per hour. AlphaSense sells primarily by seat, with reported pricing of roughly 10,000 to 20,000 US dollars per user per year and expert call credits billed separately.
Is AlphaSense an expert network or a search platform?
Both, since the Tegus acquisition. AlphaSense bought Tegus in July 2024 for 930 million US dollars, adding a library of over 150,000 investor-led expert interview transcripts covering more than 35,000 public and private companies to a corpus of roughly 450 million searchable documents. The practical difference is that AlphaSense sells you access to expert conversations that already happened, whereas GLG and Guidepoint arrange new ones. Recorded transcripts are cheaper and instant; they cannot answer a question nobody thought to ask.
Why can't an expert tell me what my customers want?
They often can describe the category accurately and still be wrong about your buyers. Compliance cooling-off periods mean the expert's information is deliberately aged, recall replaces observation, and paid experts tend to give tidier answers than real buying behaviour supports. Most importantly, an expert has no account, no invoice and no renewal date with you, so nothing they say is evidence about your specific customers' behaviour.
Can I use both expert networks and customer interviews?
That is the strongest approach. Use five to eight expert calls to build the category map and learn what to ask, then run 30 to 50 interviews with actual buyers to find out what is true of the demand side. Contradictions between expert consensus and buyer evidence are usually the most valuable finding, because a consensus every competitor also purchased is not an advantage.
Are expert network calls a compliance risk?
They carry managed risk rather than unmanaged risk. The industry operates under a framework built after the SEC's insider trading enforcement of the early 2010s, resting on real-time call documentation, MNPI surveillance and cooling-off periods for recent insiders, and the SEC's examinations division continues to focus on expert networks and alternative data. Regulated firms typically require chaperoning and pre-approval. Customer research on your own customers sits outside this regime entirely, though it carries its own privacy and consent obligations.
How many customer interviews replace an expert call?
They are not interchangeable, so the honest answer is none for questions about inaccessible knowledge, and all of them for questions about demand. A useful budgeting heuristic: one 1,500 dollar expert hour is roughly the cost of several hundred Koji credits, which is more customer interviews than most studies require. If your question is why buyers choose, spend the marginal budget on the demand side.
Talk to the demand side
Expert calls tell you how the market works. Your customers tell you what they will do. Most teams buy the first and skip the second because the second felt harder to organise.
Start free with 10 credits. Launch an AI-moderated study with your own buyers, get coded themes with traceable quotes the same day, and see how much of the expert consensus survives contact with the people who actually pay.
