When every expert agrees, ask why

Several industry experts, with different employers, backgrounds and no obvious shared interests, gave almost identical answers to the same question.

Under ordinary circumstances, that degree of agreement might be treated as strong corroboration.

Yet in a recent investment research project, the consistency itself was what made us pause.

The following case is based on an actual engagement. Companies, industries, products and other identifying details have been anonymised and generalised where necessary.

Our client was evaluating an investment in a company operating in a highly specialised industrial niche. Within that niche, a leading company was preparing to enter the public markets.

The market leader and its affiliated research institution had played an outsized role in developing industry talent. A large proportion of professionals in the field had worked, studied or maintained professional links within this broader ecosystem.

During the initial stage of our research, we interviewed industry participants from different companies and professional backgrounds.

When asked about the size and future growth of the market, almost every expert cited or endorsed the same set of projections. Based on those figures, they presented an exceptionally optimistic outlook.

The forecast closely linked this narrow market with a much larger and more visible new-energy growth narrative, producing a substantial estimate of the sector’s future potential.

The experts spoke with confidence.

Their views were also remarkably consistent across interviews.

On the surface, the market size and growth outlook appeared to have been corroborated by multiple industry participants.

But one question remained:

If ten experts repeat the same number, has that number been independently verified ten times?

The challenge to the consensus came from a different part of the value chain

I did not question the forecast because the experts lacked relevant credentials or because their accounts contradicted one another.

The opposite was true: their answers were almost too consistent.

The reason for doubt came from previous research on an upstream industry.

The niche market relied on a particular core raw material. Based on our earlier understanding of that material’s total market size and downstream application mix, this narrow use case was not one of its major destinations.

Yet the projections endorsed by the industry experts implied something very different.

When translated into the corresponding raw-material demand, the forecast would have turned a relatively limited sub-application into one of the material’s most important end markets.

Both assumptions could not easily be true at the same time.

Either the total market and application structure of the upstream material had been substantially underestimated, or the forecast for the niche market relied on assumptions that were too aggressive.

At this point, adding more experts from the same sector might not have resolved the problem.

Additional interviews could simply have produced the same answer again.

The real question was no longer how many experts supported the forecast.

It was why professionals with different backgrounds had converged on exactly the same set of numbers.

If no one appears to be lying, what should we question?

When evidence conflicts, it is natural to examine the position and incentives of the people providing it.

In this project, however, incentives alone did not offer a convincing explanation.

The experts came from different organisations and occupied different positions within the industry. Several had no obvious financial or professional interest in the companies concerned.

It was unlikely that unrelated experts had coordinated to mislead the research team.

Most were probably giving an honest account of the industry as they understood it.

That shifted the focus away from whether the experts were telling the truth and towards a different question:

Where did the “industry facts” they believed originally come from?

We began tracing how the market-size estimate had been produced and circulated.

After reviewing public disclosures, industry materials and citation chains, we found that one important source could be traced to listing-related filings prepared by an industry company, together with external market data cited in those filings.

Further tracing led to a research organisation whose independence from the industry was not sufficiently clear.

The forecast was subsequently distributed through public reporting and industry channels. It was cited by related companies, research articles and market materials.

As the same figures appeared across different formats and outlets, they gradually acquired the appearance of authoritative industry data.

Experts referred to what had already become a widely circulated market consensus.

Media and industry companies referred to corporate disclosures, expert commentary and existing research.

Search tools and AI systems then collected and repackaged the same public information, returning it through more pages and in more formulations.

An investment team might therefore encounter:

a listing-related filing;

several news and industry articles;

market descriptions published by related companies;

similar views expressed by several experts;

and an industry summary generated by search or AI tools.

These appear to be different channels.

But when traced back to their underlying sources, they may not represent multiple independent datasets. They may represent a single forecast circulating through multiple information carriers.

This does not prove that the original forecast was wrong.

It does mean that its reliability cannot be established by counting how many times it was subsequently cited.

Repetition can create the appearance of consensus. It cannot replace independent verification of the original assumptions.

Independent experts do not necessarily rely on independent sources

In investment research, we routinely consider whether interviewees have shared interests or relationships.

Experts from different companies, functions and career backgrounds are often treated as independent sources.

This project demonstrated a less visible problem.

Experts may be independent of one another as individuals while still relying on the same underlying source of information.

They may never have met and may share no commercial interests. Yet they may have read the same listing materials, absorbed the same industry narrative or relied on the same widely circulated dataset.

Evaluating an expert opinion therefore requires more than identifying where the expert works.

We also need to understand how the opinion was formed.

Did the expert directly encounter the business fact being tested?

Did the data come from the expert’s own operational, commercial or research experience, or from public industry materials?

Was the expert describing demand personally observed among customers, or repeating a forecast already accepted by the market?

What assumptions supported the expert’s expectations?

Could those assumptions be reconciled with observable evidence from raw materials, supply chains, capacity, pricing and customer procurement?

An expert interview is an important information carrier.

Whether it adds new evidence depends on whether the expert provides a new factual basis or simply restates an existing market narrative.

An expert interview first tells us what that person believes. Only by establishing why they believe it can we determine what evidence the interview has actually added.

Industry consensus is not a conclusion. It is something to be explained

Under the time pressure of a transaction, an investment team may understandably treat agreement among several experts as cross-validation.

This does not mean the team lacks professional judgement.

The research window may be only one or two weeks. Analysts may be reviewing management materials, building financial models, assessing competitors and coordinating financial, legal and other due diligence work at the same time.

Only a limited number of experts may be available within the required period. Tracing every forecast back to its original source requires additional time and research capacity.

Expert networks solve an important access problem.

They do not automatically solve the problem of source independence.

When three experts give the same answer, the key fact is not that three people agree.

The question is whether their answers were formed from three independent factual foundations.

If a customer observes increasing demand through actual procurement, a supplier sees a corresponding change in shipments, and a competitor observes the same trend through orders or capacity utilisation, the evidence may support the same conclusion from different positions.

If all three experts are citing the same market report, however, the research has collected three expressions of opinion, not three independent verifications.

Agreement becomes stronger evidence only when the paths by which the opinions were formed are also independent.

Similarly, when evidence from a different position in the value chain conflicts with industry consensus, it should not be dismissed simply because more experts support the prevailing view.

In this case, the upstream raw-material data was not automatically more authoritative than the expert opinions.

Its value was that it introduced a constraint from a different part of the value chain. That constraint exposed a contradiction in a forecast that had appeared internally consistent.

The contradiction did not prove that the market forecast was false.

It did show that the apparent consensus might depend on a common assumption that had not yet been independently tested.

Independent verification is not a vote

The purpose of independent primary research is not to find more experts who support or oppose a conclusion.

It is to identify the assumptions on which an investment judgement depends and determine where the evidence supporting those assumptions actually comes from.

Had our research remained entirely within the same specialised sector, it might have produced a highly consistent and optimistic market view.

What changed the direction of the research was placing that view back into the wider value chain and comparing it with the size and application flows of the upstream raw-material market.

The next step was not to declare that the industry forecast was wrong.

It was to break the forecast back into questions:

How had the market been defined?

Which parts of the broader new-energy demand actually belonged within this specialised segment?

How much raw-material consumption would the forecast require?

Could that demand be supported by capacity, procurement, customer projects and observable supply-chain activity?

Were the multiple datasets appearing in public materials genuinely independent?

What could the available evidence support, and what remained impossible to verify independently?

The value of independent verification is not to show that the researcher understands the industry better than its experts.

Nor is it to prove that the majority must be wrong.

Its value lies in continuing to ask where an answer originated, even when everyone gives the same answer, and whether it can be supported through a separate factual path.

Expert consensus can be the starting point of research. It cannot automatically become the conclusion of an investment decision.

The value of professional research is not in counting how many people believe the same thing, but in determining how many independent pieces of evidence those beliefs actually rest upon.