Methodology
How NarrEx scoring and verdict logic works.
This page explains the decision logic behind NarrEx outputs: what is evaluated, how figures are matched to model evidence, how verdicts are assigned, and how the score should be interpreted in practice.
What NarrEx evaluates
NarrEx reviews figures across core operating areas including growth, margins, efficiency, headcount and capital discipline. Each figure is matched against relevant model evidence so teams can quickly see where materials and model align, where they partially align, and where gaps appear.
Coverage is intentionally structured and repeatable, so two reviewers assessing the same materials arrive at consistent conclusions regardless of individual experience or time pressure.
Verdict definitions
Scoring methodology differs by workspace mode. Mode 1 (Investor) uses a 0–100 credibility score based on claim verdicts. Mode 2 (Advisor / Preparer) uses a cross-check score with a send gate based on reconciliation register status. Mode 3 (Due Diligence) uses an alignment score with risk-rated findings. The detailed verdict definitions below apply to Mode 1 — Investor workspaces. Mode 2 and Mode 3 use different label sets as summarised here.
Advisor / Preparer workspaces use Reconciled, Requires Fix, Unmapped, Definition Mismatch, Presentational Variance, and Period Mismatch. Due Diligence workspaces use Verified, Material Misalignment, Requires Further Evidence, and Not Evidenced in Model, each assigned a High, Medium, or Low risk rating.
Mode 1 — Investor verdict definitions:
Figure is consistent with the financial model within the tolerance threshold.
Directionally aligned but outside the standard tolerance band. Worth noting but not a primary concern.
Figure conflicts with the model beyond the tolerance threshold. Requires reconciliation.
No corresponding model data found for this figure.
Metric is not relevant for the selected firm type or transaction type.
How evidence matching works
NarrEx applies stricter checks to historical figures and context-aware allowances to forward-looking projections. This reflects the reality that reported actuals should be precise, while forecasts carry expected uncertainty.
When no valid model evidence is available for a figure, the system flags that figure rather than assuming alignment. Ambiguous mappings are surfaced, not silently resolved.
How to interpret the score
The scoring approach below describes Mode 1 — Investor workspaces. In Mode 2, the cross-check score reflects reconciled figures as a percentage of total figures checked, and the send gate blocks distribution until Requires Fix items are resolved. In Mode 3, the alignment score summarises overall findings alignment, with risk-rated findings driving sort order and diligence reporting priority.
In Mode 1, the score is a 0–100 directional signal summarising overall figure-to-model alignment. Higher scores indicate broader consistency between materials and underlying financial evidence. Lower scores indicate more unresolved gaps, weaker support, or missing substantiation. The score is a decision support signal, not a standalone verdict. Use it alongside qualitative judgement, market context, and other diligence inputs.
Scope and limitations
NarrEx evaluates figure-to-model alignment. It does not assess market size, competitive positioning, team quality, or any factor not represented in the submitted financial model. A high score does not imply deal quality or investment merit — it indicates that the figures in the materials are consistent with the numbers in the model. NarrEx is a decision support tool. It does not make investment recommendations.
Ready to see it in action?
Apply