Definition
The standard exists so that a SheetBrain audit report can be referenced by an author, recipient, manager, or procurement officer as evidence that a specific evaluation occurred against a specific set of rules at a specific point in time.
A spreadsheet is business-ready when it satisfies both integrity criteria and assessment criteria.
Integrity criteria
- No critical formula integrity risks.
- No unresolved structural data issues within the scanned range.
- No unresolved schema drift between the spreadsheet and prior known-good states, when prior state is recorded.
Assessment criteria
- The scan scope is explicit: sheets, ranges, and cells.
- The verdict is tied to a published, named, version-pinned detector set.
- The limitations of the evaluation are disclosed.
- The evaluation is re-runnable under the same detector set against the same source.
The verdict applies to the snapshot evaluated. A spreadsheet that was Trusted on Apr 27 may have unresolved issues on May 1 if it has been modified.
Severity classes
| Class | Meaning | Verdict consequence |
|---|---|---|
| Critical | An issue that materially compromises the correctness of the data or its derived values, such as broken references or visible formula errors in active formulas. | Critical — do not forward until fixed |
| Open | An issue that does not necessarily produce wrong values today but represents a risk to future correctness or trust, such as mixed data types, duplicate lookup keys, formula pattern outliers, or missing copied formulas. | Has open issues — not recommended for external sharing |
| Informational | An observation that does not affect verdict but may be useful to the author, such as unused named ranges, sparse columns, or hardcoded constants. | Verdict unaffected |
The exact list of detectors that map to each class is published as part of the detector set definition.
Detector sets
Every SheetBrain audit report cites the detector set used to produce its verdict.
A detector set is a named, version-pinned, immutable bundle of:
- The list of detectors active in the set.
- The severity class assigned to each detector.
- The verdict thresholds for Critical, Has open issues, and Trusted.
- A content hash that uniquely identifies the set.
- A release date.
Once a detector set version is published, it is immutable. Detectors can be added, modified, or retired only by releasing a new version. This is the mechanism that makes historical reports verifiable.
v0.1 can be re-run against v0.1 years later and produce the same verdict, provided the underlying spreadsheet has not changed. If the same spreadsheet is re-evaluated under detector set v0.4, the verdict may differ. That is expected.
The current detector set is cited directly in SheetBrain audit reports. See the published SheetBrain Detector Set v0.4 reference for the exact detector list, severity mapping, verdict thresholds, and hash. Prior v0.3, v0.2, and v0.1 detector sets remain available for reports generated under those versions.
Scope and limitations
SheetBrain checks spreadsheet structure, formulas, type consistency, and selected data-quality risks.
SheetBrain does not:
- Verify that source data is factually correct, complete, or authorized.
- Certify business logic, modeling assumptions, or the suitability of the spreadsheet for a particular purpose.
- Validate external data sources, API integrations, named-range references that resolve outside the workbook, or third-party add-on output.
- Replace the judgment of a qualified reviewer for material business decisions.
A Trusted verdict means the spreadsheet has passed a defined, reproducible technical evaluation. It does not certify that the conclusions drawn from the spreadsheet are correct.
This boundary is intentional. The standard covers what can be tested deterministically. Everything else remains the responsibility of the author and reviewer.
Report contents
Every SheetBrain audit report contains the following fields so the report can be cited, attached to a decision, or referenced in a contract without ambiguity about what was evaluated.
| Field | Description |
|---|---|
| Verdict | One of: Trusted, Has open issues, Critical. |
| Verdict timestamp | The moment the evaluation was performed. |
| Workbook identifier | Workbook name, report-scoped metadata, and stable issue identifiers used for follow-up comparison. Public reports show comparison counts, not raw fingerprint hashes. |
| Scope | Sheets and ranges evaluated, plus a count of cells scanned. |
| Detector set | Version, content hash, and detector-set label cited by the report. |
| Issued report seal | HMAC-backed metadata seal used by SheetBrain to check whether issued report metadata still matches the stored audit row. Public reports show only the seal status, not the seal input. |
| Issued metadata signature | Status-only Ed25519 signature check over issued report metadata. Public reports show only the signature status, not the signature value or manifest. |
| Audit evidence packet | Recipient-facing summary of lifecycle status, seal/signature status, detector reference, expiry, scope, lineage availability, and privacy boundaries. Public report and verifier pages also include print/PDF-friendly evidence-summary styling plus a verification footer with the report ID and verifier URL. |
| Recipient review checklist | Plain-English checklist telling recipients how to review lifecycle status, seal/signature status, detector reference, expiry, scope, findings, and limitations before relying on the audit. |
| Findings | Each finding listed by severity class, with location and explanation. |
| Limitations | The standard limitations text above, plus any scan-specific limitations. |
| Report ID | A unique, immutable identifier for the report. |
| Expiry | The date after which the report should be regenerated to reflect current state. |
Report schema v0.4 adds a status-only Ed25519 issued-metadata signature alongside the existing HMAC issued-report seal. Both checks cover report metadata such as issuer, generated time, detector-set identity, verdict/counts, expiry, and tracked fingerprint comparison counts. They deliberately exclude lifecycle changes such as revocation or supersession, so a revoked report can still verify that its originally issued metadata has not drifted. Current public reports and verifier pages also include an audit evidence packet and recipient review checklist so recipients can see the lifecycle status, seal/signature status, detector reference, scope, expiry, privacy boundary, and what to review before relying on the workbook. Browser print/save-as-PDF preserves a compact evidence summary and hides screen-only call-to-action clutter. Active audit-link rows in the SheetBrain sidebar reconcile status with the backend before display, so owners can see when a locally remembered link is active, superseded, expired, or revoked.
Versioning
This standard follows semantic versioning.
- Major versions introduce changes that affect the meaning of the verdict or the structure of reports.
- Minor versions add detectors, refine severity classifications, or clarify scope without breaking compatibility with prior reports.
- Patch versions are not used; corrections to a published version are issued as the next minor version.
The current standard version is v0.1. Versions below v1.0 are subject to revision based on feedback from real-world use. Every revision will be published; prior versions will remain accessible.
Reports generated under v0.x will state their version, and recipients should treat them as evaluations under an evolving standard.
How this standard evolves
The SheetBrain Spreadsheet Trust Standard is iterated based on:
- Findings encountered in real spreadsheet evaluations that the current detector set misses.
- Feedback from authors, recipients, managers, and procurement officers who reference reports.
- Categories of spreadsheet error that emerge as common across user populations.
- Coordination with adjacent standards in the data-quality and audit space, where relevant.
Updates are published with a changelog. Material changes to verdict semantics are versioned as a major release.
This standard is maintained by SheetBrain. It is not currently endorsed by any external standards body. Recipients should evaluate the standard's fitness for their purpose before relying on its verdicts in formal procurement, audit, or compliance contexts.
A note on intent
This standard exists because spreadsheets are routinely used to make decisions worth far more than the spreadsheets themselves, and the question of whether a given spreadsheet can be trusted has historically been answered by intuition, seniority, or hope.
A standard does not eliminate that judgment, but it gives recipients a defined surface to evaluate before applying it. Trusted under SheetBrain Detector Set v0.4 with no critical issues is a different statement from looks fine to me. Both involve human review. Only one of them is referenceable.
That is what this standard is for.