Redesign SEO audit

Audited 16 September 2026, against the local redesign working tree. This is implementation and production-build evidence; the changes are not deployed.

Coverage

All 35 indexable pages are prerendered from their real React components and content. Each returns a page-specific HTML document with one visible main H1, readable content without JavaScript, a unique title and description, one absolute production canonical, social tags and structured data. Client rendering preserves the same metadata. Navigation clears stale article tags and replaces the schema.

Group	Count	Canonical paths
Main pages	11	/, /features, /nam-rack, /ai, /download, /docs, /compare, /community, /blog, /releases, /roadmap
Legal	3	/privacy, /security, /terms
Guides	15	All /docs/<slug> entries in src/features/docs/index.ts
Articles	6	All /blog/<slug> entries generated from blogs/

The generated sitemap.xml contains exactly those canonical pages; robots.txt permits crawling and links the production sitemap. Legacy /v2, /blogs, /home, /stem-separation, /github and /contact paths retain permanent redirects. Unknown routes, guide slugs and article slugs return HTTP 404 and noindex, without a canonical or page schema. The social-card generator route /og-card is now a development-only tool and returns 404 in production.

Gaps corrected in this follow-up

• Removed the arbitrary fallback sitemap date. Known guide/article update dates remain; undated pages omit lastmod. Google recommends dates that reflect significant content changes. Google's sitemap guidance.
• Restored homepage Organization/WebSite information and added connected WebPage and SoftwareApplication entities. Removed the duplicate legacy schema from the base HTML template so page metadata has one owner.
• Connected the existing Blog index schema, retained BlogPosting author, publisher, image and date fields, and added guide/article breadcrumbs and TechArticle metadata for guides.
• Made static and runtime social-image URLs agree, including image content-hash versions. Both now expose the same actual image dimensions. The base template also references the regenerated branded social card.
• Corrected the AI search description to mention MiniMax Music 3 and Stable Audio 3 Medium as setup/availability topics, without claiming all development models are in the currently published installer.
• Prevented runtime 404 navigation from adding a canonical back to a nonindexable page. Canonical/schema suppression and sitemap exclusion consistently honor noindex.
• Removed the public social-card utility route and its production chunk; the local npm run generate-og development workflow remains available.

Verification

• Production build and strict TypeScript checks passed.
• ESLint passed with zero warnings; all 106 website tests passed.
• All ten existing mobile/desktop performance gates passed with unchanged budgets.
• The new browser regression visits all 35 sitemap pages with JavaScript disabled and enabled, verifies unique head tags and structured-data semantics, checks the actual dimensions of social assets, and compares static/runtime metadata exactly. It also tests real SPA link navigation, article-to-guide cleanup, missing pages, the internal utility URL and 404-to-home recovery.
• Existing checks cover responsive routes, crawlable internal links and anchors, image/srcset targets, old-route redirects and legal-page failure recovery.
• HTML body hashes for all 35 pages are identical before/after this SEO follow-up. Styles were not edited; no user-visible page content or layout changed.

Local evidence: output/review/seo-build.log, seo-tests.log, seo-lint.log, seo-body-comparison.json, seo-performance.json, and seo-deployed-before.json. The repeatable browser test is test/seo-pages-browser.test.mjs and runs in the existing CI suite.

Prerendering and matching canonicals follow Google's JavaScript SEO guidance. Structured data describes real content; it does not promise a rich result. Software-app review stars require genuine qualifying review/rating data, which the site does not currently have. None was fabricated. Google's software-app requirements.

Deployment and search-engine follow-up

Direct HTTP checks during this audit found the deployed production sitemap still lists 17 URLs, and /docs/keyboard-shortcuts still returns 404. The supplied PR preview's /v2/docs/keyboard-shortcuts returns 200 with the old generic HTML head and Netlify's X-Robots-Tag: noindex. This confirms that the updated local 35-page build is not the currently deployed site. Evidence is dated above; it does not describe a future deployment.

After publishing:

1. Check all 35 canonical URLs, sitemap/robots responses, old-route redirects and representative unknown URLs against the actual Netlify deployment. Verify production does not inherit the preview's noindex header.
2. Confirm apex/alternate-host and trailing-slash behavior on the CDN and inspect the resulting canonical URLs. Check social previews against the public image URLs after cache refresh.
3. Submit the canonical sitemap in Google Search Console and inspect representative homepage, guide, blog and legal URLs. Review indexing/canonical reports and run the Rich Results Test on the deployed applicable schemas.
4. Monitor Search Console page experience and field Core Web Vitals after traffic accumulates. Local lab measurements are not field data or a ranking guarantee.

Search Console ownership/settings, actual Google index coverage, rankings and external rich-result validation were not accessed or claimed verified. No changes were committed, pushed or deployed by this audit.