Organizations often say they want research that challenges assumptions. The harder question is what happens when it actually does. Consider a familiar pattern in organizational research and insights work.

A consequential problem is identified by a researcher or framed by senior leadership. A research team develops the questions, hypotheses, analytical approach, methodology, and execution roadmap. Perhaps the method involves system dynamics, predictive modeling, econometrics, machine learning, causal analysis, or another approach that is technically unfamiliar to many of the stakeholders sponsoring the work. The researchers explain the proposed approach and offer to go deeper into the methodology.

• The response is often reassuring:

• We trust the research team. We don’t need to understand all the technical details.

• The project is approved.

• Then the research begins producing evidence.

Perhaps the model identifies an unexpected causal relationship. Perhaps the estimated economic impact differs materially from leadership’s assumptions. Perhaps an initiative believed to be generating value appears less effective than expected. Perhaps an organizational capability assumed to be mature turns out not to be. Or perhaps the research exposes a systemic issue that is difficult to communicate publicly.

Something changes. Questions that seemed unnecessary before the research began suddenly become urgent.

• Why was this variable included?

• Why does the model behave this way?

• How did you calculate that number?

• Can we simplify the methodology?

• Could we look at the problem differently?

Sometimes those are exactly the questions responsible stakeholders should ask. But sometimes something more subtle is happening. The methodology was accepted before anyone knew what it would reveal. Its real acceptance is being tested only after the results have become consequential. I call this the Methodological Consent Gap.

Prospective approval, retrospective acceptance

The Methodological Consent Gap is the difference between formally approving a research methodology before execution and being genuinely prepared to accept what that methodology may subsequently reveal. At the beginning of a project, methodological trust can compensate for limited technical understanding. 

Stakeholders do not need to become system dynamicists, statisticians, econometricians, or machine-learning specialists. Organizations depend on specialized expertise precisely because no executive team can possess deep expertise in every analytical discipline. The difficulty arises when methodological distance prevents stakeholders from distinguishing among three fundamentally different reactions:

I don’t understand this result.

I don’t like this result.

There is something methodologically wrong with this result.

Those statements are not equivalent.

Yet they can become remarkably difficult to separate once research findings carry economic, political, reputational, or organizational consequences. The problem is therefore not simply a lack of technical knowledge.

It is the interaction between methodological cognitive distance and consequential evidence.

A useful way to express the underlying risk is:

Methodological cognitive distance × expectation violation × consequence of the finding

As each increases, the likelihood that methodological questions become entangled with organizational discomfort may also increase.

That is a proposition rather than an established empirical equation. But the mechanisms underlying it are well supported by several bodies of research.

The evidence-use problem is already well documented

There is no credible statistic, to my knowledge, demonstrating that a specific percentage of organizational research projects fail or change direction because stakeholders reject methodologies that generate uncomfortable findings.

Claiming otherwise would overstate the evidence.

What the literature does provide is substantial evidence for the individual mechanisms that can produce this condition.

A systematic review by Kathryn Oliver and colleagues examined 145 studies conducted across more than 59 countries on barriers to and facilitators of using research evidence in policymaking. Among the most frequently reported barriers were availability and access to research, clarity and relevance of findings, timing, and the research skills of evidence users. Collaboration and relationships between researchers and decision-makers were among the most frequently identified facilitators.

The review also reported that policymakers’ research skills and awareness were identified as barriers across dozens of studies, while personal experiences, judgments, values, political pressures, finances, and competing priorities were also found to influence the use of evidence.

FIGURE 1 — Barriers and facilitators shaping the use of research evidence 

Number of studies identifying selected barriers and facilitators to evidence use in Oliver et al.’s systematic review of 145 studies. The figures represent frequencies across the reviewed literature, not percentages of failed research projects 

This distinction matters. The problem is not merely whether an organization has evidence. It is whether decision-makers have the capability, relationships, context, and incentives required to interpret and use it.

Valuing data is not the same as being comfortable using it

A second evidence stream comes from research on data literacy.

Accenture and Qlik surveyed 9,000 full-time employees across nine countries. Although 87% recognized data as an organizational asset, only 25% believed they were fully prepared to use data effectively, and only 21% reported confidence in their data-literacy skills.

Perhaps more revealingly, only 37% said they trusted their decisions more when those decisions were based on data, while 48% reported frequently deferring to gut feeling rather than data-driven insights. Thirty-six percent said they would find an alternative way to complete a task rather than use data.

FIGURE 2 — The gap between valuing data and being able to use it 

Selected results from Accenture and Qlik’s global study of 9,000 employees. The data illustrate the difference between recognizing the value of data and feeling capable or comfortable using it 

These findings should not be interpreted as evidence that executives reject sophisticated research.

They show something more foundational: organizational enthusiasm for evidence can substantially exceed organizational confidence in interpreting it.

That gap becomes especially important when the research method itself requires reasoning about feedback loops, nonlinear relationships, uncertainty, interactions among variables, counterfactuals, probabilities, or model assumptions.

A stakeholder may genuinely trust the researcher at project approval while lacking the methodological foundation needed to independently evaluate an unexpected result months later.

Trust works easily when the evidence is unsurprising.

Its real test begins when the evidence contradicts expectations.

Organizations still struggle more with human systems than technology

The same pattern appears at the organizational level.

Wavestone’s 2024 Data and AI Leadership Executive Survey included senior data and AI leaders from more than 100 Fortune 1000 and major global organizations. In the survey, 77.6% identified culture, people, process, and organizational factors as the principal challenge to becoming data-driven, compared with 23.4% citing technology limitations.

Even after substantial improvement in 2024, only 48.1% of respondents reported having created a data-driven organization, and 42.6% reported having established a data and analytics culture.

FIGURE 3 — Becoming data-driven remains an organizational challenge 

Wavestone’s longitudinal survey shows progress in data-driven organizational maturity, while human and organizational factors remain the dominant reported challenge.

The implication is important. Organizations can invest heavily in analytics, AI, data platforms, modeling capability, and research talent while leaving the social architecture around evidence relatively underdeveloped.

That architecture includes methodological literacy, decision rights, research governance, psychological safety, incentive structures, and rules for challenging analytical conclusions. Technology can produce an insight. It cannot guarantee that an organization is structurally prepared to accept it.

What happens when the evidence conflicts with expectations?

Another body of research helps explain why methodological distance may become especially consequential once findings challenge existing beliefs. Confirmation bias—the tendency to favor information consistent with existing beliefs or prior decisions—is well established in behavioral research. An experimental study by Costa and colleagues involving 68 managers, 86 accountants, and 118 participants in a control group found evidence of confirmation bias among managers and accountants making managerial decisions. This does not mean leaders deliberately manipulate research.Confirmation bias is rarely that simple.

People can sincerely believe they are scrutinizing weak evidence when, in reality, evidence inconsistent with existing expectations is receiving a higher burden of proof than evidence that confirms them. There is also evidence for what researchers call the shoot-the-messenger effect.

Across 11 experiments, Leslie John, Hayley Blunden, and Heidi Liu found that people tend to evaluate innocent bearers of bad news more negatively. Their research suggests that negative reactions can attach not only to unwelcome information but also to the individual communicating it. Organizational communication research adds another mechanism. Fiona Lee’s work on the MUM effect examined how bad news is communicated within organizational hierarchies, finding that communicators use strategies that shape how undesirable information is transmitted. Taken together, these research streams point toward an important organizational vulnerability.

When technically complex evidence is also unexpected, consequential, and difficult to communicate, the organization is not evaluating methodology in a psychologically neutral environment. The evidence has entered a social system.

From methodological clarification to methodological erosion

None of this means stakeholders should stop questioning researchers. Quite the opposite. Good research depends on challenge. Assumptions should be tested. Models should be interrogated. Sensitivity analysis should be performed. Alternative explanations should be considered. Weak evidence should be rejected.

The crucial distinction is between methodological scrutiny and methodological accommodation. Researchers also have an obligation to translate complexity.

A system dynamics model that only another modeler can understand has limited decision value. Technical experts should be able to explain the logic, assumptions, evidence, uncertainty, limitations, and implications of their work in language decision-makers can understand. But:

Simplifying the explanation of a methodology is not the same as simplifying the methodology until its conclusions become easier to accept.

That distinction can disappear gradually.

One assumption is changed.

Then an analytical boundary moves.

Then the research question is reframed.

Then the team is asked to examine another scenario.

Then an uncomfortable result receives less emphasis.

Then a method that originally differentiated the research becomes invisible in the final communication because it is considered too complicated.

Then another iteration begins.

No single change necessarily looks unreasonable.

That is precisely why the process can be difficult to detect.

Over time, individually defensible accommodations can accumulate into what I would describe as methodological erosion: the progressive weakening or redirection of an analytical approach through repeated adjustments driven partly by stakeholder comprehension, expectations, or communication comfort rather than by new evidence or legitimate methodological concerns.

Eventually, the project may retain its original name while answering a materially different question.

Conditional epistemic acceptance

This leads to a deeper problem.

When an organization says before execution:

We trust your methodology.

but later effectively behaves as though:

We trust your methodology provided that we can understand, explain, defend, and live with what it discovers,

the organization is practicing what could be called conditional epistemic acceptance.

That condition can create powerful incentives for researchers.

Researchers learn which methods create friction.

They learn which findings take too much effort to socialize.

They learn which questions generate politically difficult answers.

They learn which analytical complexity is likely to survive executive review.

And eventually, without anyone explicitly instructing them to do so, researchers may begin selecting questions and methods partly according to what the organizational system can comfortably absorb.

That creates a dangerous possibility:

An organization may believe it has created an insights function when it has gradually created an evidence-production function constrained by organizational comfort.

The two are not the same.

A genuine insights capability exists to reduce uncertainty and discover something the organization did not already know.

If research is valuable only when it validates what the organization is prepared to hear, its discovery function has already been compromised.

This is ultimately a research-governance problem

The Methodological Consent Gap cannot be solved simply by asking researchers to communicate better.

Communication matters enormously.

But the deeper solution is governance.

Before significant research begins, organizations should explicitly address questions such as:

• What assumptions and methodological decisions have been approved?

• What evidence would legitimately justify changing them?

• Who has authority to request methodological changes?

• How will deviations from the original research question be documented?

• How will unexpected findings be challenged?

• How will methodological concerns be distinguished from disagreement with the implications?

• What findings must remain visible even if leadership ultimately chooses not to act on them?

• At what point does a pivot constitute a new research question rather than an iteration of the existing one?

For complex research, organizations may also benefit from creating methodological checkpoints before results exist. Instead of saying, “We trust the technical team,” stakeholders can invest enough time upfront to understand:

1.      what the methodology can and cannot establish;

2.      which assumptions matter most;

3.      how uncertainty will be treated;

4.      what kinds of counterintuitive results are possible; and

5.      what conditions would justify changing the analytical design.

This does not require turning executives into technical researchers. It requires sufficient methodological consent before consequence. Because once an uncomfortable number appears on the screen, methodological education is no longer occurring in a neutral environment.

The harder test of a research culture

Organizations frequently say they value curiosity.

They say they want evidence-based decisions.

They say they want researchers who challenge conventional thinking.

Those aspirations are important.

But the strength of a research culture is not demonstrated when sophisticated analysis confirms what everyone already believes.

It is demonstrated when a rigorous method produces a finding the organization did not expect—and perhaps did not want.

At that moment, the real question is no longer whether the organization values research.

It is whether the organization has built a decision environment capable of accepting what research might actually discover.

References

Accenture & Qlik. (2020). The Human Impact of Data Literacy: A Leader’s Guide to Democratizing Data, Boosting Productivity and Empowering the Workforce.

Costa, D. F., Carvalho, F. de M., Moreira, B. C. de M., & Silva, W. S. (2020). Confirmation bias in managerial decision-making: An experimental study with managers and accountants. Revista de Contabilidade e Organizações, 14, e164200. doi:10.11606/issn.1982-6486.rco.2020.164200.

John, L. K., Blunden, H., & Liu, H. (2019). Shooting the messenger. Journal of Experimental Psychology: General, 148(4), 644–666. doi:10.1037/xge0000586.

Lee, F. (1993). Being polite and keeping MUM: How bad news is communicated in organizational hierarchies. Journal of Applied Social Psychology, 23(14), 1124–1149. doi:10.1111/j.1559-1816.1993.tb01025.x.

Oliver, K., Innvar, S., Lorenc, T., Woodman, J., & Thomas, J. (2014). A systematic review of barriers to and facilitators of the use of evidence by policymakers. BMC Health Services Research, 14, 2. doi:10.1186/1472-6963-14-2.

Wavestone. (2024). 2024 Data and AI Leadership Executive Survey: Executive Summary of Findings.