A skeptic reacting to the sponsor’s thesis is not the same as independent challenge. For consequential investments, committees should build a rival explanation before discussion begins, then make the evidence discriminate between them.
By the time an investment reaches committee, one side of the argument usually has a formidable advantage.
The deal team, analyst, or portfolio manager has spent weeks building the case. It has chosen the comparables, refined the model, spoken with management or industry contacts, reconciled inconvenient data, and turned a messy set of observations into a coherent story. The investment committee sees the output of that accumulated work.
Then someone is asked to challenge it.
The imbalance is obvious. One side has a research program. The other has a meeting role.
Calling that person a devil’s advocate does not fix the problem. The challenger is still reacting to the sponsor’s evidence, terminology, causal model, and definition of the important uncertainties. Even a talented skeptic can spend the meeting debating whether the sponsor has the right answer to the wrong question.
For high-consequence investments, the stronger design is to create two independently researched cases before the committee deliberates.
The second case need not be bearish. It need not reach the opposite recommendation. It should instead offer the strongest plausible rival explanation of the same facts: a different account of what drives the asset, what the market is pricing, which evidence matters, and what would have to happen for the sponsor’s thesis to be wrong.
The committee can then do something more useful than “challenge the deal.” It can determine which observations discriminate between competing explanations.
That is a different form of dissent. And it requires a different investment-committee process.
The sponsor always has a head start
An investment thesis does more than summarize information. It determines what subsequent information means.
Suppose a private-equity team believes a target’s slowing growth is primarily a temporary sales-execution problem. Customer attrition becomes evidence of poor account management. Weak bookings become evidence of an immature sales organization. Competitor wins become a hiring problem. The facts may all be accurate, but once the causal frame is established, each new fact can be interpreted inside it.
A rival case might start elsewhere: perhaps the company is not suffering from temporary execution problems but from a weakening product advantage. The same attrition, bookings, and competitive losses now point toward a different mechanism, with very different implications for underwriting and post-close value creation.
This is why investment challenge cannot be reduced to a checklist of biases. The problem is structural: the first coherent explanation organizes the search that follows.
Research on judgment and group decision making supports that concern, although none of it should be treated as a laboratory replica of an investment committee. In a 2011 experiment involving 144 participants performing estimation tasks, Lorenz and colleagues found that exposure to others’ estimates could reduce the diversity of judgments without producing a corresponding improvement in collective accuracy. Their broader warning was that visible convergence can make a group appear more certain without making it more correct.
More directly relevant to committee design, Minson, Mueller, and Larrick found across collaborative estimation experiments that discussion could either improve or degrade judgment. A critical condition was timing: when participants first formed independent estimates, subsequent discussion could help eliminate large errors; when they did not, discussion narrowed the range of answers considered and substantially hurt accuracy in the tasks studied.
The investment implication is not that IC members should avoid talking to one another. It is nearly the opposite.
Discussion becomes more valuable when there is something independent to discuss.
If every member encounters the opportunity through the sponsor’s completed narrative before forming a view, apparent agreement after a vigorous debate may contain less independent information than the room assumes.
A new 2026 CFA Institute Research Foundation monograph makes a related argument specifically about investment committees. Bernhard Scherer describes traditional discussion-based ICs as vulnerable to speaking-order effects, hierarchy, narrative coherence, and the loss of identifiable individual views during consensus formation. He recommends, among other measures, obtaining independent judgments before consensus.
Scherer’s quantitative committee simulations should be interpreted conservatively: they use an LLM-based “narrative lab,” not observed field decisions, and the monograph itself emphasizes the difficulty of obtaining the necessary counterfactual data. But the governance diagnosis is useful. An IC cannot preserve independent information that was never elicited in the first place.
Devil’s advocacy is not independent evidence
There is a respectable intuition behind assigning someone to oppose the investment.
Dissent can expand information search. Charlan Nemeth’s work on minority influence argues that exposure to genuine disagreement can stimulate people to consider more alternatives and search more broadly, even when the dissenter is ultimately wrong.
But that literature also exposes the weakness in ceremonial devil’s advocacy: being instructed to disagree is not the same thing as arriving at disagreement through independent evidence.
In an experiment with 201 employees and managers, Schulz-Hardt, Jochims, and Frey compared groups whose members genuinely entered with different preferred alternatives against groups using an assigned devil’s advocate. Genuine preference heterogeneity was more effective at reducing confirmatory information search than the contrived dissent intervention.
Nemeth, Brown, and Rogers similarly found in experimental work that authentic minority dissent generated better effects on the thinking they measured than several versions of role-played devil’s advocacy.
These studies have important boundaries. They examine relatively constrained experimental tasks, not multibillion-dollar acquisitions or concentrated public-equity positions. And contrived dissent is not useless: other group-decision research finds that structured disagreement can sometimes improve information search and decision quality.
The practical distinction is therefore not dissent versus no dissent.
It is reactive dissent versus independently generated information.
A devil’s advocate who receives the sponsor’s 80-page memo on Monday and is told to “push hard” on Tuesday is operating downstream of the sponsor’s framing. A rival-case team that starts with the investment question and conducts its own research has at least a chance of finding a different frame.
That difference is especially important in investing because the sponsor is not merely another committee participant. The sponsor has often accumulated informational, emotional, and reputational ownership of the proposal. Walking away can mean admitting that weeks of work produced no transaction or position.
A good IC should not expect personal willpower to neutralize that asymmetry. It should design around it.
Build two explanations, not a bull case and a bear case
A two-case protocol can easily become theater of its own.
If Team A is instructed to produce the bullish case and Team B the bearish case, both can simply construct advocacy briefs. The committee receives two polished collections of selectively chosen evidence and learns little about which explanation is more likely.
The purpose is not to force symmetry. It is to produce competing causal models.
Imagine a long-only manager considering a concentrated position in a company whose margins are substantially below those of its peers.
The sponsor’s case might be:
Margins are temporarily depressed because management is investing ahead of growth. Revenue scale will reveal operating leverage, taking earnings materially above consensus.
The rival case might be:
The margin gap is structural because the company requires greater customer-acquisition and support spending than peers. Revenue growth will not produce the assumed incremental margins.
Both cases accept much of the same historical data. The disagreement is over what generates it.
The useful IC question is no longer “Are we comfortable with the margin assumptions?”
It becomes: What evidence should look different if one mechanism is true rather than the other?
Customer acquisition costs by cohort might discriminate. So might support intensity among mature customers, incremental gross margin, sales productivity after territory maturation, or the experience of former operators who have observed how the cost base behaves as the business scales.
The same logic applies in private markets. A sponsor may argue that a company is taking share because its product is superior. A rival explanation may attribute the growth to temporarily aggressive pricing or channel incentives. Rather than exchanging opinions about competitive quality, the committee can seek evidence that distinguishes product pull from commercial subsidy.
The rival case has done its job even if the sponsor wins decisively.
The objective is not balance. It is identification: finding evidence capable of telling the committee which explanation better describes the world.
That makes the second case fundamentally different from a risk section in an investment memo. A risk section asks what could damage the preferred thesis. A rival case asks whether the preferred thesis is the right explanation in the first place.
A four-step protocol for consequential decisions
The full protocol should be reserved for decisions important enough to justify duplicated effort. When it is used, four disciplines matter.
1. Separate the research paths before the narrative hardens.
Give two teams, or for smaller organizations two researchers, an identical decision question and access to the relevant evidence universe.
The sponsor develops its investment case normally. The rival path is tasked with finding the strongest plausible alternative explanation, not rebutting individual slides in the sponsor’s presentation.
Some common factual infrastructure can be shared to avoid pointless duplication. But the rival researchers should make their own decisions about which comparables matter, which experts to seek, which historical analogues are relevant, and which causal mechanism best fits the evidence.
The separation need not be absolute. Its purpose is to protect enough independence to permit a genuinely different model of the situation to emerge.
This is also where an outside view can be useful. Lovallo, Clarke, and Camerer studied private-equity investment judgments and found evidence that forming a broader reference class of analogous cases could improve forecasting relative to relying on a few salient analogies familiar to the decision maker.
The rival team should therefore be allowed to challenge not only the sponsor’s forecast, but its chosen reference class.
2. Require discriminating evidence from both cases.
Every case should identify a small number of observations for which the rival explanations make meaningfully different predictions. For each one, ask:
- If Case A is correct, what should we expect to observe?
- If Case B is correct, what should we expect instead?
- What evidence do we currently have?
- What evidence could we obtain before the decision?
- What future signal would cause us to update after investing?
This prevents the two-case process from degenerating into an accumulation of arguments.
It also changes how primary research is commissioned. Instead of asking ten experts, “How attractive is this market?”, one research path may need to test whether customers view the product as technically differentiated while the other tests whether adoption is actually driven by switching incentives. The same respondents should not automatically be used for both questions.
The goal is not expert consensus. It is information gain.
3. Elicit an independent first judgment from every IC member.
Before the group discussion begins, each voting member should record a view.
The exact instrument can vary by asset class: invest or pass; expected return; probability of thesis success; preferred position size; valuation range; or a short distribution across outcomes. Members should also identify the assumption they believe has the greatest ability to change their judgment.
The scores are not the decision. They are an information-preservation device.
The Minson-Mueller-Larrick findings make the logic concrete: independent estimates preserve a broader set of initial judgments before discussion begins compressing them. CFA Institute’s 2026 work likewise recommends having members independently assess the same case before declaring that an IC has a collective view.
For an IC chair, the dispersion is itself useful information. A 7–0 vote after discussion can look like overwhelming conviction. Seven materially different initial views that converge after the CIO speaks tell a different governance story.
4. Converge only after the alternatives are visible.
Only now should the committee attempt synthesis.
The chair’s job is not to give equal airtime to bullish and bearish speakers. It is to organize the discussion around the disagreements that could actually change the investment decision.
Where do the cases agree?
Where do they assign different meanings to the same evidence?
Which disagreement comes from different forecasts, and which comes from different objectives or risk tolerances?
Which observation could resolve the largest amount of uncertainty?
What did individual members change their minds about after hearing the competing cases, and why?
The committee should finish with one portfolio decision, not permanent philosophical pluralism. Deliberation has value: experimental research has found circumstances in which structured small-group discussion improves aggregate judgment, particularly when participants bring information or estimates worth combining.
Independence and discussion are therefore complements, not substitutes.
First preserve the alternatives. Then exploit the group.
Not every investment deserves two research teams
The strongest objection to this protocol is practical.
Duplicating research costs money. It consumes scarce senior attention. It can delay a decision. It can also manufacture false balance: when one explanation is overwhelmingly supported by evidence, an institutional requirement to construct an equal opposing case can elevate fringe arguments that deserve little weight.
Those are real costs.
A full rival-case process is most defensible when several conditions coincide: the decision is consequential; reversal will be costly; the position is concentrated; the evidence is incomplete; the sponsor has invested substantial time or reputation; and the thesis depends heavily on a narrative about the future rather than on a readily observable fact.
A platform acquisition deserves more independent challenge than a 30-basis-point liquid position.
A venture investment built around uncertain technological adoption deserves more than a small follow-on round where the critical milestones are already measurable.
A concentrated long position whose valuation depends on one non-consensus causal thesis deserves more challenge than a diversified exposure driven principally by systematic factors.
For lower-stakes decisions, lighter mechanisms are appropriate.
A premortem, asking the team to imagine that the investment has failed and explain why, can surface risks that members may otherwise hesitate to raise. An outside-view exercise can test the sponsor’s base rates. Independent member scoring can preserve disagreement at very little cost. A designated challenger can still expose neglected weaknesses.
These are useful tools.
They are simply not substitutes for an independently researched case when the decision warrants one.
The committee should spend less time debating and more time discriminating
One implication follows from the two-case design that is easy to miss.
A better IC may actually need less open-ended debate.
Investment organizations often equate a long discussion with rigorous challenge. But once rival cases have been researched properly, much of the committee’s work can become more precise.
The sponsor believes retention will improve because implementation problems are being fixed. The rival believes attrition reflects a product gap.
What measure distinguishes them?
The sponsor believes a market slowdown is cyclical. The rival believes the category is approaching saturation.
What leading indicator should differ?
The sponsor believes management guidance is conservative. The rival believes management is extrapolating a temporarily favorable cohort.
Which historical or cross-sectional comparison would separate those interpretations?
This style of discussion also makes post-investment learning more useful. The original cases and their discriminating predictions can be revisited after six or twelve months. The committee does not have to reconstruct what “the thesis” was from a retrospective narrative.
It can ask which model predicted the evidence that subsequently arrived.
That matters because decision processes are otherwise remarkably good at rewriting history.
McKinsey research on 1,048 major corporate decisions found a strong association between decision-process practices, including explicit consideration of contradictory viewpoints and major uncertainties, and decision outcomes. The study was survey-based and observational, so it does not establish that those practices caused better performance. But its central implication remains relevant: strong analysis does not guarantee a strong decision process.
IC design deserves the same seriousness firms apply to financial modeling.
Where Enquire fits: create independence before asking for synthesis
The operational difficulty in a two-case protocol is obvious: independent research is expensive precisely because it duplicates search.
That is where research infrastructure can help, provided it preserves rather than erases differences between the cases.
Enquire’s current capital-markets offering combines structured research with expert perspective and explicitly supports teams in identifying where views diverge, pressure-testing investment narratives, and surfacing assumptions that may be mispriced. Its product also supports gathering responses from multiple vetted experts and AI-led asynchronous interviews.
For a rival-case process, the useful application is not asking one system to generate a bullish paragraph and a bearish paragraph.
It is maintaining separate research paths.
One path can investigate the sponsor’s causal explanation and seek evidence that would validate or falsify it. Another can research a rival mechanism, drawing on different operators, customer perspectives, industry specialists, or adjacent-market evidence. The resulting expert perspectives can then be compared to see where independent sources converge and where the divergence itself deserves further investigation.
AI can make it cheaper to maintain those parallel searches. It should not prematurely merge them.
The committee benefits from synthesis only after it has preserved something worth synthesizing.
The purpose of dissent is not disagreement
Investment committees do not need more people performing skepticism.
They need a process that gives a plausible alternative enough time, evidence, and independence to become genuinely informative.
Sometimes the rival case will overturn the sponsor’s thesis. More often, it may strengthen the investment by identifying the precise assumption that needs additional diligence, changing the price at which the risk becomes attractive, reducing position size, or establishing a clearer monitoring trigger after investment.
And sometimes it will lose completely.
That is not wasted work.
A committee that considers two serious explanations and finds one overwhelmingly better supported has learned something quite different from a committee that begins with one explanation, challenges it vigorously, and survives the challenge.
The first has compared models of the world.
The second has tested the sponsor’s skill at defending one.
For the investments that matter most, the question for the IC chair should therefore come before anyone enters the room:
What is the strongest independently researched explanation under which this investment is a mistake, and what evidence would let us tell the two cases apart?
Sources and further reading
- Bernhard Scherer, Investment Committees: Governance and Design Choices, CFA Institute Research Foundation rpc.cfainstitute.org
- Julia Minson, Jennifer Mueller, and Richard Larrick, “The Contingent Wisdom of Dyads,” Management Science pubsonline.informs.org
- Stefan Schulz-Hardt, Marc Jochims, and Dieter Frey, “Productive Conflict in Group Decision Making” psych.uni-goettingen.de
- Charlan Nemeth, “Minority Influence Theory,” UC Berkeley irle.berkeley.edu
- Jan Lorenz et al., “How Social Influence Can Undermine the Wisdom of Crowd Effect,” PNAS pnas.org
- Dan Lovallo, Carmina Clarke, and Colin Camerer, “Robust Analogizing and the Outside View,” Strategic Management Journal sms.onlinelibrary.wiley.com
- Dan Lovallo and Olivier Sibony, “The Case for Behavioral Strategy,” McKinsey mckinsey.com
- Gary Klein, “Performing a Project Premortem,” Harvard Business Review hbr.org
- Enquire, Capital Markets enquire.ai