When Sources Disagree: What Your Strategy Deck Should Say

Show both sources, name the divergence, and state what your recommendation assumes. This article explains why expert interviews and documents so often disagree, how to weigh the competing sources, and a step-by-step conflict register that keeps every contradiction visible in the deck instead of silently resolved in the appendix.

When expert interviews and data-room documents point in opposite directions, your strategy deck should neither average the numbers nor quietly pick a favourite. It must state both findings on the affected slide, explain why they diverge, and declare what the recommendation assumes. Disagreement between primary research and corporate records is not a research failure; it is normal evidence that requires structured evaluation.

In strategy engagements, contradictory evidence typically arises from three structural friction points rather than conscious dishonesty:

• Definitional drift: Corporate documents and industry experts often use identical terms for different metrics, such as confusing logo retention with net revenue retention, or measuring total company averages against a specific tier.
• Temporal mismatch: Data-room documents reflect historical reporting periods with inherent reporting lag, whereas an interview captures real-time operating conditions and emerging shifts.
• Positional knowledge and incentive bias: Former executives recall conditions from their tenure, operating managers report segment-level lived experience, and external advisers often repeat second-hand market narratives as established facts.

In Psychology of Intelligence Analysis, Richards J. Heuer Jr. notes that analysts construct mental models that filter incoming information, often leading them to overlook inconsistencies.^[1] When conducting expert interviews, treating divergence as diagnostic data rather than an analytical nuisance is the first step toward sound strategy.

To resolve conflicting information rigorously, strategy practitioners can adopt proven methodologies from national security intelligence analysis. Rather than relying on intuition or seniority, source evaluation should follow structured criteria: temporal recency, proximity to the operational fact, independence from other reporting streams, and whether an assertion is directly observed or inferred.

Intelligence tradecraft provides three foundational frameworks for evaluating contradictory material:

• The Admiralty code: Used across NATO intelligence bodies, this system evaluates source reliability (graded A-F) separately from information credibility (graded 1-6)^[2]. Under this two-character matrix, a completely reliable source (grade A) can still provide information that is only possibly true (grade 3), preventing analysts from confusing source seniority with factual accuracy.
• ICD 203 analytic tradecraft standards: Intelligence Community Directive 203 requires analysts to properly describe the quality and credibility of underlying sources, distinguish raw information from assumptions, and explicitly express uncertainties associated with major judgments^[3].
• Analysis of Competing Hypotheses (ACH): Developed by Richards Heuer, ACH uses a matrix to test how evidence relates to alternative explanations, focusing on disproving hypotheses because evidence consistent with multiple outcomes lacks diagnostic value.

By evaluating the witness and the data independently, you prevent persuasive interviewees from overriding audited financial records, while ensuring that stale reports do not blind the engagement team to genuine market inflections.

The conflict register is a systematic protocol that converts evidentiary contradictions into explicit, auditable governance decisions. When an analyst identifies a contradiction during research, the finding is logged and processed through four discrete operational steps before any slide is drafted.

1. Log the contradiction: Record the disputed claim alongside exact source citations for both sides, including interview transcript timestamps, document titles, version numbers, and page references.
2. Map positional authority: Detail what each source is structurally positioned to observe, noting potential gaps in access, reporting horizons, or commercial incentives.
3. Identify the root cause of divergence: Determine whether the contradiction stems from mismatched metrics, differing timeframes, geographic scope, or unverified inference.
4. Assign a formal disposition: Categorise the conflict into one of four mandatory outcomes: Reconciled, Weighted, Open, or Tested.

The four conflict register outcomes establish clear rules for how the analysis proceeds:

• Reconciled: The divergence dissolves once definitions, scopes, or baseline dates are normalised across sources.
• Weighted: One source is judged superior based on verifiable proximity or methodology; the preferred figure forms the baseline, while the alternative is noted on the slide.
• Open: The contradiction cannot be resolved with current data; both figures are shown on the slide, and the strategic recommendation explicitly states its underlying assumption.
• Tested: A targeted verification task is commissioned before final sign-off, such as an additional data cut, customer call, or technical review.

Two mandatory rules govern this process: no evidentiary conflict may be resolved silently by an analyst, and every Open or Weighted conflict that affects a strategic conclusion must appear on the relevant slide, as outlined in a robust human review process.

To illustrate how the conflict register functions in practice, consider an illustrative hypothetical scenario from a software due diligence engagement. During primary research, a former sales director states that enterprise customer churn is accelerating sharply (Interview Transcript 4, minute 14). Conversely, the vendor data room contains a customer retention spreadsheet indicating that enterprise churn has remained completely flat over the preceding eight quarters (Data Room, Folder 3, Workbook 'Retention_v2.xlsx', Tab 'Enterprise').

Instead of choosing between the senior witness and the spreadsheet, the engagement manager logs the issue in the conflict register:

Because the data room measured customer logo counts while the sales director observed seat reductions and discounting within top accounts, the conflict dissolves into a coherent operational reality. The slide narrative is rewritten accordingly:

Draft wording: 'Enterprise logo churn is flat annually; however, net revenue churn in tier-one accounts is expanding, driven by seat contractions. The core recommendation assumes an account expansion and pricing remediation, rather than a customer defection crisis.'

If the data room export had measured both logo and revenue retention and still shown flat metrics, the disposition would become Open. The deck would present both data points side by side, explicitly stating that the investment case assumes the audited revenue figures hold, while highlighting account monitoring as a key post-close milestone.

A central question for strategy teams is where conflicting findings should reside. Burying evidentiary discrepancies in an appendix is a governance failure that exposes boards to unquantified operational risks. When an unresolved contradiction touches a key recommendation, it belongs directly on the main slide in clear, transparent prose.

Structuring conflicting findings on executive slides requires precise presentation:

• Place the tension directly alongside the finding: State the primary conclusion, cite the baseline source, and include a concise note detailing the alternative data point and its origin.
• Define the sensitivity: Explain how the strategic recommendation would change if the alternative evidence proves correct, framing the uncertainty objectively.
• Identify decisive indicators: Specify the exact trigger metrics or operational signals that would invalidate the baseline assumption and prompt a strategy revision.

Executive committees and boards value the transparency of an acknowledged conflict over the false comfort of smoothed data. When presenting to a sceptical board, showing that your team has surfaced, tested, and bound an ambiguity builds credibility and protects decision-makers from unexpected blind spots.

As strategy practices incorporate artificial intelligence into document ingestion and research synthesis, tools must adhere to strict analytical tradecraft. An AI platform must not merely summarise individual files in isolation or hallucinate a consensus between mismatched sources; it must actively identify contradictions across primary and secondary inputs.

Any enterprise strategy tool supporting complex analysis should meet three core operational criteria:

• Cross-corpus contradiction detection: The platform must cross-reference primary interview transcripts against uploaded financial models and corporate contracts, automatically flagging divergences in scope, numbers, or sentiment.
• Bidirectional source traceability: Every claim, figure, and flagged contradiction must provide a direct, one-click link to the underlying document passage, table cell, or interview transcript timestamp.
• Persistent auditability: To ensure that an AI analysis is reproducible, the system must preserve the frozen evidence set and underlying reasoning chains so judgments can be audited months later.

While advanced algorithms can efficiently surface contradictions and map evidentiary links, assigning the final disposition across Reconciled, Weighted, Open, or Tested categories remains a human responsibility. Just as structured analytic techniques are designed to assist analysts rather than replace their judgment^[1], final strategic accountability rests with named authors.

An AI-native strategy consulting platform should be engineered to enforce rigorous evidence standards throughout the advisory lifecycle. The platform structures strategy deliverables directly from uploaded client documentation and primary research streams, ensuring that strategic arguments are grounded in auditable facts.

To support the conflict register protocol, Decisity cross-checks primary research findings against the internal documents in an engagement, surfacing contradictions with both sides of the evidence in view and providing direct links back to original sources. Every claim, number, and recommendation generated within the platform is fully traceable to its source with an end-to-end audit trail, and this source-tracing persists so that strategic decisions can be re-examined months later against the exact evidence base upon which they were built.

From an architectural and data governance standpoint, Decisity is designed to align with the EU AI Act and is engineered in line with ISO 27001 and SOC 2 control principles. Client documents and confidential data rooms are never used to train shared machine learning models, and all data is hosted under European data residency standards, providing strategy leaders with enterprise-grade security alongside rigorous analytical integrity.

Related reading:

• Correcting a wrong number in a board paper after the meeting
• Presenting AI-Assisted Strategy Analysis to a Skeptical Board

• US Government: A Tradecraft Primer: Structured Analytic Techniques for Improving Intelligence Analysis (March 2009)
• Pangea Research: The Admiralty Code, the NATO 6x6 source evaluation system
• Office of the Director of National Intelligence: Intelligence Community Directive 203, Analytic Standards