---
title: Verify the result
description: A minimum acceptance protocol for research artifacts.
reviewedOn: 2026-09-08
sourceRefs: [evidence-lab-methodology]
---

## Result

The artifact receives an explicit `accepted`, `rejected`, or `needs correction` decision against criteria written before the run. Mechanical checks and domain judgments are recorded separately.

| Check layer       | Typical question                                            | Evidence                                         |
| ----------------- | ----------------------------------------------------------- | ------------------------------------------------ |
| Completeness      | Are all required files and sections present?                | File list, schema, or checklist                  |
| Fidelity          | Do claims, quotations, dates, and numbers match sources?    | Source locations and comparison notes            |
| Execution         | Did scripts, queries, formulas, or exports run as intended? | Logs and deterministic test output               |
| Presentation      | Does the final artifact work in its target application?     | Rendered inspection and screenshots where useful |
| Domain acceptance | Is the interpretation acceptable for the research purpose?  | Named expert decision and caveats                |

## Actions

1. Freeze the candidate artifact and the acceptance checklist so the reviewed version is unambiguous.
2. Check file completeness, required formats, naming, and internal structure.
3. Compare key claims and transformations with the permitted primary material or strongest available source.
4. Check dates, names, numbers, links, quotations, citations, and identifiers as separate high-risk fields.
5. Run available schemas, tests, formulas, link checks, extraction checks, or other deterministic validation.
6. Open or render the final artifact in the target application and inspect the real reader path.
7. Record findings by criterion, corrections, limitations, residual risk, reviewer, and final decision.

## Check

Each criterion has a result and evidence location. The reviewed artifact version matches the delivered version. The record names who performed mechanical review and who accepted the scientific or domain judgment; unresolved findings are visible in the decision.

## If it fails

Do not mark an unverifiable criterion as passed. Replace it with an observable condition, obtain the missing source or reviewer, or label it as a pending manual expert judgment. After correction, re-run the original criterion against the new artifact version.

**Next:** [Independent use](/en/quality/independent-use/).