A narrative can be internally coherent — every slide consistent with the next, every number flowing cleanly from the model — and still be wrong.
This is one of the central challenges in investment diligence and it is one that existing platforms are poorly equipped to address. Data platforms surface information. Deal rooms organise documents. CRM systems track pipeline. None of them test whether the internal logic of a narrative is proportionate to the evidence that underpins it.
Coherence is a property of the story. Confirmation requires external evidence. The distinction matters more than it might appear.
The coherence trap
When a management team presents investment materials, they are not presenting raw data. They are presenting an interpretation of data — a story about what the numbers mean, why the trajectory will continue and why the assumptions are reasonable. That story is constructed to be coherent. Every element supports every other element. The growth rate justifies the headcount plan. The headcount plan explains the burn rate. The burn rate is covered by the raise. The raise is sized by the milestones. The milestones are driven by the growth rate.
The loop closes. The narrative is internally consistent. And that consistency can feel like evidence.
But internal consistency is not the same as external support. A model can be technically coherent — every formula referencing every other formula correctly — while embedding assumptions that are not grounded in observable evidence. A narrative can flow smoothly from problem to solution to traction to projections while each step relies on a premise that has not been tested.
Coherence is a property of the story. Confirmation requires external evidence.
Where the gap appears
The coherence trap is most dangerous in the middle of a diligence process, not the beginning. At the start, scepticism is high. Questions are broad. Nothing has been accepted yet. But as the process progresses, each answered question reduces apparent uncertainty. The narrative becomes more familiar. Internal discussions normalise the key assumptions. By the time the investment committee meets, the story has been repeated so many times that its premises feel established.
This is not because investors are careless. It is because coherence is genuinely useful as a heuristic. A story that holds together is more likely to reflect reality than one that contradicts itself. The problem is that coherence is a necessary but not sufficient condition for credibility.
What confirmation actually requires
Confirmation requires tracing each important claim back to observable evidence — not to other claims. Revenue growth is confirmed when it is supported by a revenue model that contains the drivers capable of producing it. Margin expansion is confirmed when the cost structure assumptions that underpin it are present in the model. Customer acquisition claims are confirmed when the model contains CAC figures consistent with the sales capacity and spending assumptions it also contains.
NarrEx is designed around this distinction. The system does not evaluate whether a narrative is coherent. It evaluates whether each claim is supported by independent evidence from the financial model. A claim that appears coherent within the narrative but lacks model support is flagged as Unsupported. A claim where the narrative and the model directly conflict is flagged as Contradicted.
The score does not measure story quality. It measures alignment between assertion and evidence.
The practical implication
The coherence trap suggests a specific question worth asking in every IC discussion: which of the key assumptions in this narrative have been confirmed by independent evidence and which have only been confirmed by other assumptions in the same narrative?
That question is hard to answer informally. It requires tracing the evidential chain for each claim — not just checking whether the story holds together, but checking whether each link in the chain is attached to something real.
That is the problem NarrEx is built to solve.
A morning-room failure mode
Picture a familiar IC rhythm: the deal lead summarises the company in confident sentences borrowed from the deck. Partners nod because the summary is smooth. Someone asks a sharp question about competition; the answer is also smooth. The model is referenced only for multiples and scenario outputs. Two hours later, the committee approves a thesis nobody has traced claim-by-claim to independent evidence.
Nothing in that room feels negligent. It feels professional. That is what makes the coherence trap insidious: professionalism and coherence correlate in real life, so the room mistakes narrative polish for evidentiary depth.
Breaking the pattern requires a procedural interrupt—not more intelligence, but a different order of operations. Before discussing valuation, ask for ten minutes of “confirmation mapping”: name the three claims that matter most, then name the artefact outside the narrative stack that supports each one. Silence during that exercise is data.
Confirmation without cynicism
Seeking confirmation is not the same as presuming bad faith. Most management teams believe their story. Confirmation work simply asks whether the belief is tethered to objects investors can re-check next quarter: a model line, a cohort cut, a hiring table, a customer contract sample, a usage metric tied to revenue.
When confirmation is thin, the right emotional stance is curiosity, not suspicion. “Help us see what you see” opens doors. “Prove it” closes them. The intellectual task is identical; the social outcome is not.
Why coherence still matters
None of the above argues against wanting a coherent narrative. Incoherent stories are often wrong stories. The claim is narrower: coherence is the beginning of diligence, not the end. A coherent story plus thin confirmation should receive a different confidence label than a coherent story plus thick confirmation.
Institutions that internalise that distinction make fewer category errors. They still miss outcomes—markets are uncertain—but they miss for better reasons. They confuse luck with skill less often. They reward partners who surface gaps early rather than partners who tell the smoothest story.
If you take one habit from this piece, make it the interrupt: when a narrative feels finished, ask what would still be true if you deleted every slide except the model and the data room. Whatever remains is the confirmation core. Everything else is packaging—often excellent packaging, but packaging nonetheless.
Micro-example: two versions of the same quarter
Imagine a company that beats revenue plan while gross margin slips slightly. Version A of the story: “We invested ahead of growth; efficiency returns next half.” Version B: “Mix shifted to enterprise pilots with higher services burden; margin normalises as attach rises.” Both can be coherent. Neither is confirmed until you see hiring timing, services revenue recognition and margin bridges in the model that match the chosen version.
Coherence lets both stories sound credible in the room. Confirmation asks which version is actually embedded in the forecast mechanics. Teams sometimes discover they were telling Version A verbally while Version B wandered into the slide footnotes. That is not lying; it is misalignment. Catching it early prevents a board meeting six months later where everyone argues about which story was “official.”
Organisational roots: who owns the chain of evidence
Coherence problems worsen when no one owns cross-artefact consistency. Sales owns the deck narrative; finance owns the model; product owns the roadmap. Without a named integrator, each function can be honest locally while the composite is misleading globally.
Sophisticated firms assign a deal desk or strategic finance function to reconcile materials before external release. Early-stage companies rarely have that luxury. Investors can simulate the role by requesting a single owner to sign off that deck claims map to model lines—not perfect, but a forcing function.
When coherence and confirmation align beautifully
The best materials feel almost boring: every dramatic sentence points to a table, every table points to a definition, every definition stayed stable for several reporting cycles. Excitement lives in the market opportunity, not in rhetorical compression. Those packs are rare, but they leave committees calmer because the work of doubt has already been done collaboratively.
Seeking that standard is not hostility to storytelling. It is respect for storytelling’s power. Stories move capital; evidence governs it. Coherence is not confirmation—but when both arrive together, you are looking at a culture that takes external trust seriously.