Understanding Your Audit Report
A plain-language guide to the six report sections, presentation, score, limits, and supporting evidence.
Every completed audit opens on a plain-language Summary. Start with the result and first recommended change, choose work in What to fix, then check the supporting proof in Evidence. A Presentation button opens an eight-page version of the same saved audit for review or PDF export.
The Six Report Sections
Start here. It shows the plain verdict, saved brand mentions and website citations, the first recommended change, evidence completeness, four score areas, and important limitations.
The saved answer tests: which AI services mentioned the brand, made it a primary suggestion, or credited the audited website. Missing individual test rows are never turned into invented rates.
A ranked now/next/later plan with expected effect, effort, ownership, supporting evidence, and a verification step. Estimates remain planning guidance until a like-for-like re-audit confirms movement.
The page checks and finding cards behind the result. Technical brand-consistency and structured-data diagnostics are collapsed as an advanced section.
Directional differences from earlier saved page audits, uncovered topic groups, and an estimated audit-gap breakdown. Page, depth, date, or report version may differ, so use this to find questions—not to claim traffic, revenue, market share, or a winner.
The supporting proof and run details: saved questions and answers, crawl observations, classifications, citations, request summaries, limitations, and the privacy-filtered JSON export.
Score Ranges
Every completed check and the overall audit use a 0–100 diagnostic rubric. These bands are fixed thresholds, not population percentiles, market position, or a business-outcome forecast. The audit score uses completed checks and checks completed with limits. Compare two reports only when the page, scope, enabled checks, and evidence mix are alike. If no score was saved, the report shows “Not scored”—never an F or zero.
Issue Severity Levels
Completed modules can emit specific issues. Each stored issue uses one of the five schema severity values below, plus its category, evidence, proposed fix, ownership, impact, and effort when available. Severity is an audit prioritization label, not proof of business impact.
The audit observed a condition that can invalidate or materially block the tested path, such as an inaccessible page or missing crawl inputs. Verify the evidence first, then investigate immediately.
A substantial gap in an enabled check, such as blocked named crawlers or a missing sitemap. It is usually a strong candidate for prioritization after critical findings.
A meaningful improvement opportunity that does not invalidate the audit. Confirm relevance to the submitted page and business context before scheduling it.
A contextual or lower-impact observation. It may be useful, but it should not displace verified critical, high, or medium work solely to improve the score.
Descriptive context that may require no action. Keep it with the receipts, but do not treat it as a defect merely because it appears in the issue list.
Estimated Score Changes
Some qualifying issues may include an estimate: “if you fix this, the audit score could change by X points.” The system compares the current measured page signals with a simulated change.
An estimate is planning guidance, not a guaranteed score increase or business result. Its source and evidence quality remain visible. Verify the change by re-auditing the same page under the same conditions.
The Evidence Section
The sixth section contains the supporting proof. It keeps saved individual records and totals separate from the plain-language conclusion, and warns when a total has no matching individual test rows. Check important claims here before acting on them.
Receipts can expose the following data when the corresponding capabilities ran. If supporting data is unavailable, the report labels that separately from an empty result:
The view renders saved evidence, subject to table-specific query limits. It can include provider prompts and responses, citation URLs, classifier output, crawl signals, screenshots, and source-waterfall entries. A missing table is treated differently from an empty result.
Behind the report, loadAuditEvidence(auditId) loads the evidence groups and counts in parallel. If a data table is unavailable, the report marks that evidence as unavailable rather than pretending the result was empty.
The Eight-Section Presentation
Choose Presentation in the report header to open ?view=deck. The eight sections cover the plain verdict, four score areas, advanced brand checks, AI answer results, compared pages, ranked actions, and evidence limitations. The browser view is responsive on mobile; “Print / Save PDF” uses fixed A4 pages. Legacy editorial and dossier links resolve to this canonical presentation rather than a conflicting report format.