Content Coverage Audit Method.
Coverage is not the amount of text a system publishes. It is the verified relationship between a declared requirement universe and the content, evidence, connections and document responsibilities observed inside a bounded corpus.
An audit measures a declared model—not an imaginary perfect page.
The requirement universe is fixed before the corpus is scored. Each requirement carries a definition, scope, evidence rule, importance class and intended document owner.
Content coverage auditing is the controlled comparison of a bounded content corpus against a versioned requirement universe to determine which concepts, entities, attributes, relations, questions and evidence obligations are supported, partial, unsupported, misplaced or absent.
Presence is recorded, but supported coverage requires the declared meaning and evidence condition.
Page-level gaps and corpus-level gaps are different analytical objects and must remain separate.
The correct repair may be a sentence, section, relation, consolidation, new page or deliberate exclusion.
One subject can be present and still remain uncovered.
Coverage is audited across five non-interchangeable layers. Passing a lower layer does not automatically satisfy the layers above it.
Mention coverage
The required expression or resolved equivalent appears inside an eligible content surface.
Concept coverage
The content communicates the required meaning with enough context to distinguish it from adjacent concepts.
Evidence coverage
The claim or explanation carries the evidence type required by the audit model.
Relation coverage
The requirement is connected to the entities, attributes, causes, consequences or processes needed for its role.
Ownership coverage
A responsible document or section clearly owns the requirement while supporting surfaces route toward it.
From requirements to repairs through an inspectable decision chain.
A coverage percentage is the end of several methodological decisions, not the beginning. Every stage freezes an output needed to reconstruct the final result.
Audit objective
Define the system decision the audit must support.
Requirement universe
Register concepts, entities, relations and obligations.
Eligible corpus
Freeze pages, sections, languages, states and date.
Detection rules
Specify what counts for every requirement and layer.
Evidence capture
Record exact passages, structures and relationships.
Coverage states
Assign supported, partial, absent and other states.
Ownership map
Identify primary, supporting and competing surfaces.
Gap register
Separate content, evidence, relation and architecture gaps.
Repair sequence
Order gaps by obligation, impact and dependency.
Audit manifest
Freeze model, matrix, decisions, limits and provenance.
Four operations. No empty console states.
Each operation exposes its input, rule, output, evidence requirement, review condition and prohibited inference. Without scripting all protocols remain visible; with scripting they become an interactive instrument.
Define what the bounded corpus is expected to cover
Determine whether observed content satisfies the requirement
Classify what is missing before proposing a repair
Route each requirement to the surface that should own it
Requirements define the denominator. Evidence defines the intersection.
The synthetic set model prevents the audit from confusing “everything published” with “everything required.” Only eligible requirements with qualifying evidence enter supported coverage.
Counts toward supported coverage under the current model.
Enters diagnosis; it is not automatically one kind of gap.
Recorded separately so useful information is not erased.
Unknown and absent items remain in the denominator when applicable.
One requirement. One state. Evidence attached.
This synthetic matrix shows how several requirements can appear inside a corpus while producing different valid coverage states.
| REQ ID | REQUIREMENT | LAYER | OBSERVATION | EVIDENCE | OWNER | STATE | ACTION |
|---|---|---|---|---|---|---|---|
| REQ-001 | Define the primary entity | Concept | Canonical definition in lead section | Direct passage | /entity-a/ | SUPPORTED | Retain |
| REQ-002 | Distinguish entity from category | Concept | Both named but difference unexplained | String presence only | /entity-a/ | PARTIAL | Add distinction |
| REQ-003 | State evidence origin | Evidence | Claim visible; origin unavailable | No traceable source | /method/ | UNSUPPORTED | Attach provenance |
| REQ-004 | Connect entity to process | Relation | Both appear on page | No explicit predicate | Unassigned | INFERRED | Validate relation |
| REQ-005 | Answer eligibility question | Question | Direct answer and conditions | Passage + scope rule | /eligibility/ | SUPPORTED | Retain |
| REQ-006 | Provide regional rule | Context | Outside declared geography | Boundary record | N/A | NOT APPLICABLE | Exclude denominator |
| REQ-007 | Declare update date | Freshness | No visible or recorded date | Collection log | /entity-a/ | UNKNOWN | Verify state |
| REQ-008 | Canonical document ownership | Ownership | Three pages compete equally | Routing map | Conflicted | CONFLICT | Resolve responsibility |
Eight gap types require eight different repairs.
A useful audit diagnoses the missing function, not merely the missing phrase.
Presence gap
An eligible requirement has no qualifying observation in the bounded corpus.
Meaning gap
The subject is mentioned but its required definition, distinction or context is incomplete.
Evidence gap
A claim appears without the evidence type or provenance demanded by its requirement.
Relation gap
Required elements exist but the connection, predicate, direction or sequence is missing.
Intent gap
Information exists but does not satisfy the user task or decision the document should support.
Ownership gap
A requirement is scattered, duplicated or orphaned without a responsible canonical surface.
Freshness gap
Coverage once existed but its validity window, source state or update obligation has expired.
Conflict gap
Eligible surfaces support incompatible treatments that remain unresolved or undisclosed.
Repair sequence follows obligations and dependencies—not visual drama.
The synthetic queue shows why an evidence gap can outrank a larger wording gap and why an ownership conflict may need resolution before new content is written.
Priority is a bounded decision, not a universal score.
TAO priority combines the requirement’s declared obligation, decision impact, dependency position, evidence risk and repair effort. The components remain visible; no composite score replaces reviewer judgment.
Critical dependencies are repaired first. Cosmetic completeness waits behind errors that can propagate into several downstream pages or claims.
Six measures. Six limited questions.
Scores are reported by layer and requirement class. A single blended number is never allowed to hide why a corpus passed or failed.
Presence rate
observed requirements ÷ eligible requirementsShows detectability only; mentioned or contradicted items may still count as observed.
Supported coverage
supported requirements ÷ eligible requirementsCounts only requirements that satisfy their declared meaning and evidence rules.
Partial coverage
partial requirements ÷ eligible requirementsIdentifies incomplete treatment without merging it into fully absent coverage.
Relation coverage
supported required relations ÷ eligible required relationsRequires an explicit predicate, direction or sequence—not simple co-occurrence.
Ownership completion
requirements with valid owners ÷ ownership-eligible requirementsExposes orphan, competing and misplaced document responsibility.
Critical gap rate
unsupported required items ÷ required eligible itemsSeparates mandatory failures from contextual opportunities and optional depth.
Four audits. Four different units of coverage.
All values are synthetic. The examples demonstrate decision logic rather than claims about live websites or external systems.
Eight shortcuts that manufacture false completeness.
Each failure changes the numerator, denominator or meaning of the result and therefore belongs inside the audit record.
Keyword substitution
A detected phrase is counted as complete treatment of its requirement.
Corpus-derived universe
The requirement set is built only from existing content, hiding every true absence.
Duplicate inflation
Repeated treatment across pages is counted as additional distinct coverage.
State collapse
Partial, inferred, contradicted, absent and unknown are converted into one binary state.
Denominator editing
Difficult requirements disappear after observation to improve the reported rate.
Page-corpus collapse
A requirement absent from one page is called absent from the complete content system.
Repair overproduction
Every gap becomes a new page even when consolidation or a small clarification is correct.
Authority inflation
An internal coverage rate is presented as proof of external trust, ranking or recognition.
Every percentage should be reconstructable.
The manifest connects each reported value to its requirement model, corpus snapshot, evidence decisions and gap actions.
Every method node. One controlled research route.
MTH/08 diagnoses coverage and gap states. MTH/09 then tests whether the evidence supporting those states is valid, relevant, independent and traceable.