The difference between an ad hoc expert call and a structured sector intelligence program is the difference between a transaction and a system. A transaction produces a data point. A system produces a thesis.
Top hedge fund research teams have internalized this distinction. They do not run expert calls when they need to fill a gap. They run sector intelligence programs that are continuously operating — structured pipelines that produce compounding insight rather than isolated data points.
The counterintuitive finding: a 12-expert sector program, run with proper structure, produces more actionable investment intelligence than 40 calls run without architectural discipline. Volume is not the edge. Structure is.
This guide walks through the architecture of a high-performing 12-call sector intelligence program: how it is designed, how the interview guide is built from the thesis outward, how synthesis cadence prevents drift, and what the final output looks like when it reaches the investment committee.
“The quality difference between a structured expert program and an ad hoc call series is not incremental. It's categorical. The structured program produces a thesis. The ad hoc series produces notes.”
Program Architecture: The 12-Call Structure
A 12-call program is not 12 random experts. It is a deliberate sampling of the information landscape — a structured representation of every perspective that bears on an investment thesis. The allocation of calls is as important as the calls themselves.
The recommended call allocation for a 12-expert sector program breaks down as follows:
Three calls with former company insiders — management-level alumni of the target company. These calls establish the baseline operational picture: unit economics, competitive positioning, management cadence, and historical decision patterns.
Three calls with direct competitors — former operators at key competitors in the sector. This cohort provides the competitive intelligence layer: how rivals perceive the target's strengths and weaknesses, where they see market share shifting, and what competitive dynamics are underappreciated by the consensus.
Three calls with ecosystem participants — customers, distributors, and suppliers. This cohort anchors the thesis in commercial reality. Distribution economics, customer retention dynamics, and supplier leverage ratios are best understood through direct ecosystem participants, not company IR.
Two calls with sector context experts — former analysts, consultants, and industry observers who have mapped the sector from an elevated vantage point. These calls provide the structural framing that prevents thesis drift — the baseline of what is already widely understood versus what would constitute a genuine variant perception.
One bear case call — someone with a publicly known or professionally held negative view on the sector or company. This call is explicitly structured to surface the strongest arguments against the thesis, not to be won.
This allocation is not arbitrary. Each cohort has a distinct signal type. The insider cohort produces operational fidelity. The competitor cohort produces competitive signal. The ecosystem cohort produces commercial ground truth. The sector context cohort produces structural framing. The bear case call produces adversarial pressure-testing. A program that over-indexes on insiders and ignores ecosystem participants will produce a distorted picture regardless of call quality.
“The bear case call is the most important call in the program. It's also the hardest to schedule, because nobody wants to spend time proving their thesis wrong.”
— Portfolio manager, long/short equity fundThe sequencing of calls within the program also matters. The recommended sequence is: sector context experts first (to establish the structural baseline), then insiders and competitors in parallel (to build the operational and competitive picture), then ecosystem participants (to pressure-test commercial assumptions), and the bear case call last — when the team has enough conviction to be meaningfully challenged by it.
The Intelligence Brief: Designing the Interview Guide
The interview guide is not a general-purpose questionnaire. It is built outward from the investment thesis — each component of the thesis generates a set of questions designed to stress-test that specific component.
The construction process starts with the thesis map. For each thesis component — for example, "channel economics are favorable and sustainable" — the research team designs two to three questions that would either confirm or disconfirm that component. The questions are not written to confirm the thesis. They are written to expose its weakest assumptions.
Question hierarchy is the most overlooked element of interview guide design. The standard error is to open with specific, thesis-confirming questions that anchor the expert to the interviewer's frame before the expert has revealed their own. The corrected approach uses a two-stage hierarchy:
Stage one: open-ended questions that reveal the expert's mental model before any thesis-specific framing is introduced. Ask the expert to describe how the distribution model works, what the key margin drivers are, or what they view as the primary risk factors in the sector. Let the expert's frame emerge unprompted.
Stage two: thesis-specific questions that test specific thesis components using the expert's own framing where possible. If the expert described gross margin as driven by three variables in stage one, the stage two question can ask directly about those three variables — which moves faster, which is most controllable, which is the one the market is misunderstanding.
The practical distinction: a bad question is "Do you think the company's margins are sustainable?" — it is a yes/no question that anchors the expert to a single variable and invites confirmation. A good question is "What are the key variables that determine gross margin in this distribution model?" — it reveals the expert's mental model first, which makes every subsequent answer more interpretable.
“We rewrote our interview guide template three times. The third version was the one that stopped asking experts to confirm what we already believed.”
The interview guide should also be specific to expert cohort. The guide used for a former company insider focuses on operational detail — unit economics, internal decision-making, the gap between what was communicated externally and what was known internally. The guide used for an ecosystem participant focuses on commercial behavior — purchase volumes, switching costs, pricing dynamics, and what has changed in the last twelve to eighteen months.
Synthesis Cadence: Updating the Thesis in Real Time
The most common failure mode in a multi-week expert program is deferred synthesis. The research team runs eight or ten calls, takes detailed notes, and then attempts to synthesize everything at the end — at which point the earliest calls are no longer fresh, contradictions between early and late calls are difficult to reconstruct, and the thesis has drifted without anyone noticing.
The corrected cadence is rolling synthesis: a structured synthesis session held after every three to four calls, before the next cohort is engaged. The session should be fixed in the calendar at program start — 45 minutes, the full research team, a defined agenda.
The synthesis session agenda has three components. First, a structured review of new claims from the week's calls — each claim is logged against the thesis component it bears on, and its consistency with prior claims is assessed explicitly. Second, an update to the conviction score with documented reasoning — not a gut check, but a specific statement of which new claim or contradiction moved the score and why. Third, identification of the highest-priority open question for next week's calls — the question that, if answered well, would produce the largest update to the thesis.
The rolling thesis document is the operating record of the synthesis process. It is updated synchronously during the synthesis session — not drafted afterwards from notes, which introduces recency bias and allows the sharpest disagreements to be softened in retrospect. The conviction score is tracked on a zero-to-ten scale, with each score change documented alongside the specific claim or contradiction that drove it.
The indicators that synthesis is working are behavioral: the interview guide changes between weeks based on what the program has learned; the conviction score moves in either direction; the team's framing of the bear case evolves after engaging with it directly. If none of these are occurring, synthesis is not happening — the team is accumulating notes, not updating a thesis.
One underappreciated benefit of rolling synthesis is that it makes the program adaptive. A three-week program that synthesizes weekly will often redirect its fourth and fifth cohort calls based on what was learned in the first two cohorts. A program that synthesizes only at the end has no mechanism for this adaptation — it runs the full call sequence and then discovers, too late, that the most important question was never asked.
The conviction score discipline is worth treating as a team norm rather than an individual practice. When the full research team agrees on the score and the reasoning that produced it, the investment committee conversation is fundamentally different. The team arrives not with a narrative, but with a documented conviction trajectory.
From Program to Investment Decision
The final output of a well-run 12-call sector intelligence program is not a research memo. A research memo narrates what experts said. The structured investment brief that should emerge from a disciplined program is fundamentally different in structure and purpose.
The structured investment brief has five required components.
One: the updated thesis statement — a single paragraph that reflects the thesis as it stands at program close, not as it was stated at program open. If the thesis has not changed at all over twelve calls, that is itself a finding worth noting.
Two: the claim register summary — the top ten highest-conviction claims from the program, each attributed by EXP identifier and cohort type, each tagged with a confidence level. This is the evidentiary backbone of the brief.
Three: the contradiction log — unresolved contradictions between expert claims, with an explicit statement of why each contradiction remains open. A contradiction log that is empty is almost always incomplete, not evidence of perfect consistency.
Four: the bear case summary — the strongest arguments against the investment thesis, drawn primarily from the bear case call and any contradictory claims surfaced during the program. The bear case summary is presented in its most compelling form, not as a strawman.
Five: remaining open questions — what the program did not resolve, with a note on whether the unresolved questions are answerable through additional expert engagement or represent genuine epistemic limits on the thesis.
The investment committee should receive this brief, not a narrative summary of what experts said. The brief enables the committee to interrogate the thesis directly — to challenge the contradiction log, to pressure-test the bear case summary, to ask why specific open questions remain unresolved. A narrative summary does not support this kind of interrogation; it invites passive reception.
The format discipline also has an archival benefit. When the investment ultimately plays out — whether the thesis is confirmed, disconfirmed, or complicated by events — the structured brief is a precise record of what was known, what was uncertain, and what was wrong at the point of decision. This is the raw material of institutional learning, which compounds across programs.
Structured sector intelligence programs compound over time in a way that ad hoc call series cannot. A fund that runs six well-structured programs per year for three years has built a proprietary sector intelligence archive — annotated claim registers, resolved and unresolved contradiction logs, updated thesis trajectories — that represents genuine institutional knowledge.
This archive is genuinely difficult to replicate. Not because the inputs are proprietary — expert networks are available to every fund — but because the structure, the synthesis discipline, and the institutional memory that accumulates from running the process well are organizational capabilities, not access advantages.
The inputs are not secret. The structure is the edge.