> ## Documentation Index > Fetch the complete documentation index at: https://docs.gradial.com/llms.txt > Use this file to discover all available pages before exploring further. # QA Reports > Understand what Gradial's QA reports surface, how to read them, and how to integrate quality checks into your content workflow. ## Why This Matters Quality review is the step that gets skipped when the launch date is tomorrow. Not because anyone thinks it's optional — because it's manual, it's slow, and it's the last thing standing between a finished page and a deadline. So it gets sampled: check a few pages, assume the rest are fine, ship. Then the disclaimer is missing on the one page a regulator looks at, or the brand name is styled three different ways across a campaign, or the alt text is empty on the hero of the highest-traffic page on the site. QA reports run on every output rather than a sample, before anything reaches a human reviewer — so review becomes a decision, not an inspection. *** QA reports evaluate content against the rules your organization has defined — surfacing failures, explaining why they occurred, and guiding resolution before content reaches reviewers or goes live. Rules are configured in the [Rules Manager](/docs/marketing-brain/managing-skills). Once rules are active, QA runs automatically on every task output. *** ## Reading a QA Report When a QA review runs, Gradial generates a report showing how content performed against each active rule. ### Report summary At the top of every report: | Field | What it shows | | - | - | | **Rules failed** | Count of rules the content did not pass | | **Rules passed** | Count of rules the content met | | **Incomplete** | Checks that require manual verification | ### Result detail Each rule result includes: | Field | Description | | - | - | | **Title** | The rule name | | **Type** | Whether this is a brand/compliance Rule or an Accessibility check | | **Status** | Passed, Failed, or Incomplete | | **Explanation** | What was found and why it failed | Click any result to expand the full explanation — including the exact content that triggered the failure and guidance on how to fix it. ### Status definitions * **Passed** — Content meets the standard * **Failed** — A violation was detected that must be corrected * **Incomplete** — Automated testing found something that requires manual verification before a final determination can be made *** ## Issue types QA reports surface issues across five categories: | Category | What it catches | | - | - | | **Brand & editorial** | Voice and tone violations, unapproved terminology, messaging inconsistencies | | **SEO** | Missing or duplicate meta descriptions, heading hierarchy problems, keyword usage | | **Compliance** | Missing required disclosures, unapproved claims, regional regulatory language | | **Accessibility** | Alt-text, heading structure, ARIA, color contrast — see [Accessibility Reports](/docs/quality-assurance-and-accessibility/accessibility-report) for full detail | | **Functional** | Missing asset references, broken links, incomplete metadata | Not all issues carry equal weight. Compliance and accessibility failures have regulatory implications; editorial issues may be lower priority. Use the severity and explanation fields to triage. *** ## QA in your content workflow The authoring agent produces output guided by active rules. Planning-agent rules apply early, before authoring begins. Integration-agent rules apply during content creation. After the authoring agent completes work, the QA agent evaluates the output against all active rules for that workspace and integration. Failed and incomplete checks appear in the report with explanations. Content with failures is flagged before it reaches a human reviewer. Fix the flagged issues and re-run QA to confirm. Content with a clean report moves to review with confidence. With baseline quality validated automatically, reviewers can focus on strategic messaging, creative direction, and stakeholder alignment — not catching compliance issues. *** ## Scaling QA across teams QA reports are most valuable when distributed teams — regional marketers, agency partners, extended business teams — are creating content without deep familiarity with every brand guideline. **The report explains, not just flags.** Each failure includes the reasoning: what the rule requires, what was found, and how to fix it. Authors self-correct before submitting for review, which reduces back-and-forth. **Pass rates improve over time.** As authors encounter and resolve the same rule types repeatedly, the failures decrease. Teams that track pass rates over time see measurable quality improvement without increased review overhead. **Give the trend a home.** A per-output report answers "is this page ready"; it doesn't answer "how are we doing." Ask Grady for a [custom app](/docs/insights/custom-apps) to get a dashboard across QA results — pass rates by team or campaign, the rules that fail most often, and what's still waiting on a fix. **Add rules as patterns emerge.** If the same issue keeps appearing and no rule currently catches it, add one. The rules library grows in response to real production patterns. *** *To configure the rules that power QA reports, see [Rules & Skills](/docs/marketing-brain/managing-skills). For accessibility-specific results and WCAG remediation guidance, see [Accessibility Reports](/docs/quality-assurance-and-accessibility/accessibility-report).* *** ## Who this is for | Role | What Gradial handles | | :- | :- | | **Web Producer / Content Author** | Issues caught and explained before review, not after a stakeholder finds them | | **Brand Manager** | Every output checked against brand standards, not a sample | | **Legal / Compliance Reviewer** | Required disclosures and regulated wording verified on every page, with a record of the check | | **Accessibility Lead** | WCAG conformance measured across the estate rather than spot-checked | | **Marketing Leadership** | A launch that doesn't depend on whether anyone had time to review it properly | ### How this maps to your function In a regulated environment the cost of a missed disclosure isn't a correction — it's exposure. Sampling is the standard practice because full review doesn't scale manually, which means the assurance you have is statistical rather than actual. Checks that run on every output change what you're able to claim. See [Regulated Industries & Compliance Workflows](/docs/governance/regulatory-compliance). Brand drift is rarely a decision. It's the accumulation of small inconsistencies introduced under deadline — the name styled two ways, the CTA that doesn't match the pattern, the tone that slipped. Consistent automated checks make the standard real at the point content is produced, rather than something audited quarterly and corrected retroactively. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.