Evidence Verification · 28 July 2026 · 8 min read
Confidence Classification in Evidence Reporting
By Digital Asset Claims Research Desk·Investigation Team
- Evidence Standards
- Reporting
- Confidence Classification
Every finding in a digital asset investigation report carries a different degree of evidential support. Confidence classification is the discipline of making that difference explicit, rather than presenting every conclusion in the same confident register regardless of how well it is actually supported.
A report that states its uncertainty honestly is more useful to its recipients than one that sounds confident throughout but cannot support that tone uniformly.
Defining the Classification Tiers
Verified fact applies to findings that are directly observable from a primary source and independently checkable by anyone with access to the same source material.
Corroborated inference applies to conclusions supported by multiple independent sources that together make the conclusion strongly probable, without any single source proving it outright.
Unverified lead applies to material that has not yet met the corroboration standard but is retained in the case file because it may warrant further investigative attention.
Recording Classification at the Point of Entry
Classification is assigned when a finding is first written into the case file, alongside the specific evidence supporting that classification, rather than being decided later during report assembly.
This point-of-entry discipline prevents the classification from being influenced by how the overall narrative is shaping up, keeping the assessment tied to the evidence available at that moment.
A documented rationale accompanies each classification, explaining specifically why the finding meets the standard for its assigned tier and not a higher one.
Preventing Language Inflation During Drafting
Reports go through multiple drafting passes aimed at clarity and readability, and this process carries a real risk of gradually overstating findings if classification is not anchored explicitly throughout.
An explicit classification system requires editors to check each revised sentence against its assigned tier, preventing cautious original language from drifting into unwarranted certainty.
This check is applied as a distinct editorial step, separate from proofreading for style, specifically because language and evidential accuracy require different kinds of review.
Updating Classification as Evidence Changes
Investigations are ongoing processes, and a finding's classification may change as new corroborating evidence emerges or as an earlier source is found to be less reliable than initially assessed.
Every classification change is documented with a timestamp and a stated rationale, preserving a full audit trail of how the case file's assessment evolved over the course of the investigation.
This living-document approach ensures that a report reflects the most current, well-founded assessment of the evidence at the time it is issued, rather than an earlier and potentially outdated view.
Confidence classification tiers used in case reporting
| Tier | Standard required | Example |
|---|---|---|
| Verified fact | Directly observable, independently checkable | A specific on-chain transaction |
| Corroborated inference | Multiple independent sources aligning | Shared infrastructure linking two platforms |
| Unverified lead | Single source, not yet corroborated | A single forum mention of an address |
| Reclassified finding | Updated following new evidence | A lead upgraded after independent confirmation |
Frequently asked questions
Who decides how a finding is classified?
The analyst responsible for the finding assigns an initial classification against defined standards, and significant classifications are subject to review by a second analyst before inclusion in a final report.
Can a finding be downgraded after initial classification?
Yes. If a source is later found to be less reliable, or an alternative explanation emerges, the classification is revised downward and the change is documented with its rationale.
Why not simply exclude unverified leads from a report?
Unverified leads can still be useful to a recipient deciding where to direct further investigation, provided they are clearly labelled as unverified rather than presented as established fact.
Does confidence classification slow down report production?
It adds a documentation step, but it produces a more defensible and genuinely more useful report, which generally saves time overall by reducing the risk of later disputes over what was actually established.
Confidence classification is a discipline that keeps a report's language honest relative to its underlying evidence. By defining clear tiers, recording classification at the point of entry, and updating findings as evidence evolves, a report communicates exactly what has been established, what has been inferred, and what remains an open lead.
