R006 — Propagated Errors
Executive Summary
Key Takeaways
- ✓ R006 flags any cell whose last calculated value is an Excel error value — formula cells and hardcoded cells alike, since both states represent a model that is not calculating cleanly.
- ✓ Every R006 finding carries high confidence, because whether a cell currently evaluates to an error is a deterministic fact already present in the workbook, not an inferred pattern.
- ✓ R006 is critical severity and one of five critical-override rules — a single triggered finding caps the model's letter grade below A or B regardless of the numeric score.
- ✓ An error confined to one cell rarely stays confined — because formulas reference other formulas, one error value typically propagates into every downstream cell that depends on it.
Purpose¶
R006 detects any cell in the workbook whose last calculated value is an Excel error — a formula that is
currently returning #REF!, #VALUE!, #DIV/0!, or one of the other standard Excel error values. Unlike
rules that infer a pattern across a row, R006 checks a single, directly observable fact per cell: does this
cell's cached value indicate an error.
Detection Logic¶
R006 evaluates every node in the workbook's formula graph — every formula cell and every hardcoded cell alike — and flags any cell whose recorded error state is set. One finding is produced per affected cell. Because error state is read directly from the workbook rather than inferred from a comparison against other cells, this rule requires no minimum sample size and no dominant-pattern threshold: a single cell in error is sufficient to trigger a finding.
Why It Matters¶
An error confined to a single cell rarely stays confined. Because formulas commonly reference other formulas, one cell's error value typically propagates through every downstream calculation that depends on it, directly or indirectly — a single broken lookup on an assumptions tab can silently turn into dozens of downstream error values, or, if wrapped in error-masking logic elsewhere in the model, into a silently substituted default value that looks like a legitimate number. R006's fact — this cell currently evaluates to an error — is unambiguous in a way few other findings in the engine are, which is part of why it is one of five rules capable of capping a model's overall letter grade under the SM-2.0 Scoring Methodology, regardless of how high the model's numeric score otherwise is.
Evidence Produced¶
A triggered R006 finding's evidence identifies the specific sheet and cell address of the error, the exact error value the cell evaluates to, and a description stating that the cell currently evaluates to that error — a direct, observable fact rather than an inferred judgement. See the Finding Data Model & Confidence page for the full evidence and provenance model this finding is built on.
Severity¶
Critical. Weight 9.0. R006 is one of five critical-override rules under SM-2.0 (alongside R001, R002, R004, and R023): if it triggers even once, the model's overall letter grade is capped below A or B regardless of the numeric score the weighted formula produces. Confidence on every R006 finding is high, since error state is a deterministic fact rather than an inferred pattern.
Example¶
Trigger. A formula on a Debt Schedule worksheet reads =Revenue!C12/Revenue!C13, and the referenced
denominator cell currently evaluates to zero, so the cell's cached value is #DIV/0!. R006 flags this
cell directly: its recorded value is an error state.
Non-trigger. The same formula, but the denominator cell holds a nonzero value, so the division resolves to an ordinary numeric result with no error state. R006 does not fire, regardless of whether the result itself is otherwise reasonable — that judgement belongs to other rules and to manual review, not to R006.
Recommended Remediation¶
The root cause of the error is investigated and corrected at its source — the broken reference, the divide-by-zero condition, or whatever upstream fault produced the error value — rather than suppressed with error-masking logic (a distinct pattern the engine addresses separately). Once the source is fixed, every downstream cell that depended on the error value should be re-checked, since a fixed source can change previously error-masked or stale downstream values.
Related Rules¶
R006 shares the Structural taxonomy category with R001, R002, and R004, each published in this same initial phase of the Rule Reference; the remaining Structural-category rules follow in later phases and are indexed on the Rule Reference page as they are published.
Related Articles¶
- Formula Error Types in Financial Models — the Knowledge Centre's buyer-facing reference to the principal categories of formula error, including the visible error values this rule detects.
Version History¶
No version-specific changes to R006's severity, weight, or category are recorded as of engine version 2.1.0. This section will be updated in the same edit as any future change to the rule's declared attributes.
How OXXON tests thisRun a free structural check with FMAE
Frequently Asked Questions
What does FMAE rule R006 check for?
Any cell in the workbook whose last calculated value is an Excel error — #REF!, #VALUE!, #DIV/0!, #NAME?, #N/A, #NULL!, or #NUM! — regardless of whether the cell itself is a formula or a hardcoded value.
Why is R006 confidence always high?
Because whether a cell's last cached value is an error is a directly observable, deterministic fact read from the workbook — there is no pattern-matching or judgement involved, unlike rules that infer a row's dominant pattern.
Does R006 flag hardcoded cells too, not just formulas?
Yes. R006 evaluates every cell in the formula graph regardless of whether it currently holds a formula or a typed value, because both a broken formula and an error value typed in as a placeholder represent the same underlying fact — the model is not calculating cleanly at that cell.
Why is a single error value treated as critical?
Because formulas commonly reference other formulas, a single error value frequently propagates into every cell downstream of it, so one error at the source can invalidate a large portion of the model's outputs — the reasoning behind R006's inclusion among the five critical-override rules, documented on the SM-2.0 Scoring Methodology page.
Related Articles
R001 — Hardcoded Cells
R001 (Hardcoded Cells) flags a formula cell that has been overridden by a typed numeric value, in a row where the surrounding cells otherwise share a clear, dominant formula pattern. It is critical severity, carries the highest weight of any FMAE rule (15.0), and is one of five critical-override rules under the SM-2.0 scoring methodology — meaning a single R001 finding caps the model's overall letter grade below A or B regardless of its numeric score.
R002 — Broken Links
R002 (Broken Links) flags a formula cell whose reference has broken — pointing at a deleted named range, a deleted worksheet, a reference-error cell, or an external workbook link the parser could not confirm resolves. It is critical severity and one of five critical-override rules under the SM-2.0 scoring methodology, meaning a single R002 finding caps the model's overall letter grade below A or B regardless of its numeric score.
R004 — Formula Inconsistency
R004 (Formula Inconsistency) flags a formula cell whose logic differs from the dominant pattern the rest of its row otherwise follows — an extra term, a different function, or a shifted reference. It is the general case of R001, catching a different formula rather than a typed constant, and deliberately excludes hardcoded cells from its own sample so the two rules never double-count the same cell. R004 is critical severity and one of five critical-override rules under the SM-2.0 scoring methodology.
FMAE Rule Taxonomy
Every rule in the FMAE structural rule pack declares a category attribute at the point it is defined in source — this is not a classification imposed on the rules afterward for documentation purposes, it is the classification the engine itself uses. Six categories cover all 26 rules — Structural (18 rules), Assumptions Governance (1), Integrity Controls (2), Structural Hygiene (1), Aggregation Logic (1), and Model Governance (3). This page publishes that taxonomy as the FMAE equivalent of a control catalog's classification scheme, cross-linked to the Rule Reference page for each member rule.
FMAE Scoring Engine — SM-2.0 Methodology
FMAE computes a model's risk score as 100 minus the triggered rules' combined weight, normalized against a fixed basis, currently 207.0 under the active SM-2.0 methodology. SM-1.0, the engine's original scoring basis (170.0, covering R001–R021), is retained as a frozen historical reference rather than deleted from the record. Five rules — R001, R002, R004, R006, and R023 — are critical-override rules. If any of them triggers, the resulting letter grade is capped so a model cannot be graded A or B regardless of how high its numeric score is. This page documents the exact formula, the versioning between SM-1.0 and SM-2.0, and a fully worked example.
Finding Data Model & Confidence
Every FMAE finding carries the same fixed structure — which rule triggered it and that rule's identity at the time it ran, its evidence, a narrative explanation, and a confidence label of high, medium, or low. Confidence is a deterministic categorical label derived from how strongly a row's cells agree on a pattern, not a machine-learning probability estimate. Stating this plainly matters, since describing FMAE's confidence as AI-derived or probabilistic would overstate what the engine actually does.
Formula Error Types in Financial Models
Formula errors in financial models fall into four principal categories: visible error values (including #REF!, #VALUE!, #DIV/0!, #NAME?, #N/A, #NULL!, and #NUM!), which display in cells and are immediately apparent; silent formula errors, which produce plausible-looking values but incorrect results; structural formula errors, which arise from incorrect model construction rather than incorrect values; and logic errors, which occur when a formula correctly implements an incorrect financial relationship. Each category requires different detection methods and carries different risk implications.