Conflicting expert views do not automatically mean the evidence is weak. They may reflect different facts, definitions, incentives, horizons, geographies, or causal models. Before averaging the answers or selecting the one that fits the thesis, research teams should determine why the disagreement exists and what evidence could resolve it.
Ten expert interviews produce an awkward result. Six respondents expect industry pricing to remain resilient. Three expect material deterioration. One argues that list prices are no longer the right measure because rebates and contract terms are changing faster.
The easiest synthesis is a vote count. It is also usually the least informative.
Consensus can hide the useful evidence
Investment research has strong incentives to compress. Portfolio managers do not want ten transcripts, and analysts need a conclusion that can enter a model. The output becomes a range, an average, or a sentence saying that most experts agree.
That compression can erase the mechanism behind the answers. Six experts may agree because they share the same outdated data. Three may disagree because they operate in a customer segment that is turning first. The outlier may be the only respondent measuring the variable that matters.
Identify what differs
Most disagreements fall into a manageable set of categories:
Facts: respondents have different or differently dated information
Definitions: they use the same word for different measures
Assumptions: they answer under different unstated conditions
Roles and incentives: their positions shape what they see or emphasize
Horizons: a near-term operational view conflicts with a long-term structural view
Geography or value chain: each respondent accurately describes a different part of the market
Causal models: they accept the same facts but disagree about what those facts will produce
The category determines the follow-up. A factual gap needs fresher evidence, a definition gap needs a clearer question, and competing causal models need a discriminating test.
Ask what would separate the models
The best follow-up is rarely, Why do you disagree with the others? That encourages debate and can push the expert toward defending a position.
Ask instead:
What are you observing that leads to this view
Under what conditions would your answer change
What would we see next if your explanation is correct
Which customer, geography, or channel should behave differently
What evidence would make the competing explanation more credible
These questions turn disagreement into research design. They identify observations that should look different depending on which explanation is right.
Build a disagreement ledger
A compact ledger should sit between the raw interviews and the investment memo. For each material disagreement, record:
The precise question
The competing answers
The reason each expert gives
Relevant differences in role, geography, horizon, or information
Evidence shared by both sides
Evidence that would discriminate among the views
The next research step or signpost
Do not average incompatible answers
Aggregation can help when experts estimate the same defined quantity independently. It is much less useful when respondents are answering different questions or describing different parts of a system.
Weight answers using relevant experience, information access, calibration, and reasoning quality. But do not delete a materially different explanation solely because it receives less weight.
The right output may be conditional rather than singular. Pricing may remain resilient in enterprise accounts while deteriorating among smaller customers. Near-term demand may be strong even as the long-term market structure weakens.
Protect independence before synthesis
Expert evidence becomes less useful when later respondents are exposed to an emerging consensus. Interview guides narrow, follow-up questions converge, and social influence reduces independent information.
Keep initial responses separate where possible. Use a consistent core question, capture the reasoning before sharing other views, and only then ask experts to react to a competing explanation. This preserves both independent judgment and the value of direct challenge.
Where Enquire fits
Research systems often compress multiple expert perspectives into one polished answer too early. Enquire can help preserve alignment and divergence across traditional calls, asynchronous AI-led interviews, and shorter expert inputs while retaining the context behind each view.
The useful output is not an automated consensus. It is a structured record of where experts agree, why they differ, which assumptions carry the disagreement, and what the team should test next.
A strong synthesis explains the disagreement
Expert disagreement is not always profound. It may reflect poor sampling, ambiguous questions, or stale experience. But when credible respondents differ for identifiable reasons, the disagreement is data.
The strongest synthesis does more than report the majority view. It explains which parts of the market support each answer, what evidence would distinguish them, and how the investment conclusion changes under each case.
Sources and further reading
1. M Granger Morgan, Use and Abuse of Expert Elicitation in Support of Decision Making for Public Policy, PNAS; 2. Jan Lorenz et al, How Social Influence Can Undermine the Wisdom of Crowd Effect, PNAS; 3. Abigail Colson and Roger Cooke, Expert Elicitation Using the Classical Model to Validate Experts Judgments; 4. Philip Tetlock et al, Forecasting Tournaments Tools for Increasing Transparency and Improving the Quality of Debate; 5. Charlan Nemeth, Minority Influence Theory; 6. Enquire, Capital Markets