Improve SEO

[Musilah]

Severity Levels

These are the Severity Levels of each SEO checklist change:

• P0 — Critical: Must be correct before production launch or major content release.
• P1 — High: Should be part of every SEO rollout and backlog for all active sites.
• P2 — Recommended: Strong improvements that raise quality, trust, and long-term discoverability.

P0 — Required On Every Indexable Page

1. Every page must have a unique title, description, canonical URL, and H1

• Unique <title> for every indexable page.
• Unique meta description for every indexable page.
• Canonical URL pointing to the final production URL, always using HTTPS.
• One clear H1 that includes the page's real topic, product, or target keyword.
• Homepage H1 must contain actual brand/product terms, not only slogan language.

Rules:

• Never point canonicals to ngrok, localhost, preview or dev domains.
• Never reuse the homepage title/description on product pages or docs index pages.
• Never let /docs, /about, /product, and / all share the same H1 structure.

---

2. Every page must ship production-safe social metadata

Checklist:

• og:title, og:description, og:url, og:image
• twitter:card, twitter:title, twitter:description, twitter:image
• Image should be stable, production-hosted, and ideally 1200x630.
• og:url must match the canonical URL.
• Use the correct og:type for the page where practical:
  • website for generic pages
  • article for articles/docs posts if you maintain article metadata
  • product for product pages

Rules:

• Never use preview-host image URLs in social metadata.
• Never inherit homepage OG values on all child pages.

---

3. Every page must include valid, page-appropriate structured data

• JSON-LD is one of the clearest machine-readable trust signals for Google and AI systems.
• Broken or misleading schema is worse than no schema.

Minimum schema expectations:

• Homepage:
  • Organization
  • WebSite
  • SoftwareApplication or Product where applicable
• Product pages:
  • Product
  • BreadcrumbList
• Documentation/article pages:
  • TechArticle or Article
  • BreadcrumbList
• FAQ-heavy pages:
  • FAQPage

Rules:

• Validate that JSON-LD actually renders as JSON, not a literal string template.
• Use real production URLs in schema.
• Only claim implemented features and supported protocols.
• Only emit datePublished, dateModified, and similar fields if they will be kept accurate.
• Only keep SearchAction if the target is a real, crawlable search experience.

Practical standard:

• Every indexable page should have at least one relevant schema block.
• Sitewide entity schema should connect with @id consistently

4. Every page must be internally connected and free of dead references

Why it matters:

• Removed pages that remain in nav, sitemap, llms files, or internal links waste crawl budget and hurt trust.
• Docs and product ecosystems need deliberate internal linking for discoverability.

Checklist:

• No internal links to removed pages.
• No sitemap entries for removed pages.
• No llms.txt or llms-full.txt references to dead URLs.
• Every page should link to at least one related page:
  • product to docs
  • docs to product
  • homepage to strongest commercial or technical asset

Rules:

• If a page is removed, remove or redirect every reference to it.
• Redirect old high-value URLs if they have backlinks or are cited externally.

---

P1 — Required On Every Site

5. Every site must have a clean crawl-control package

Why it matters:

• Search engines and AI crawlers need a consistent discovery layer.
• The most common org-wide failures are conflicting robots rules, missing sitemap entries, and missing AI-readable summaries.

Checklist:

• robots.txt
• sitemap.xml
• llms.txt
• llms-full.txt where useful for richer AI ingestion
• Correct sitemap declaration in robots.txt
• Preview/non-prod environments set to noindex or blocked appropriately

AI crawler policy:

• If llms.txt exists, the robots strategy should not contradict it.
• Avoid a situation where the site advertises AI-readable content and then blocks GPTBot, ClaudeBot, PerplexityBot, or Google-Extended by mistake.

---

6. Every site must have a reliable sitemap strategy

Why it matters:

• The sitemap is not just a list of URLs; it is a canonical crawl plan.

Checklist:

• Include all important indexable pages.
• Exclude test pages, preview pages, dead docs pages, and temporary routes.
• Match trailing slash behavior exactly with canonical behavior.
• Prefer lastModified only if you can maintain it accurately.
• If lastModified is noisy or wrong, omit it rather than lying.

Rules:

• Avoid known low-value fields if they are not useful in the setup.
• Do not keep URLs in the sitemap that redirect unnecessarily.
• Do not let sitemap URLs conflict with canonical URLs.

---

7. Every site must enforce a production-only deployment baseline

Why it matters:

• Several audit failures were caused by development builds, preview URLs, or dev-mode assets being exposed publicly.
• Search systems treat these as quality and trust regressions.

Checklist:

• Production domain serves a real production build.
• No HMR/devtool/Turbopack dev assets on public URLs.
• No preview-domain canonicals.
• No preview-domain OG images.
• No staging or ngrok URLs in metadata, schema, RSS, or structured content.

Rules:

• Public production domains must never run next dev.
• Preview URLs should be noindex.

---

8. Every site must have a trust and legal baseline

Why it matters:

• E-E-A-T depends on visible ownership, legitimacy, and contactability.
• This matters especially for B2B infra, open-source platforms, and hardware products.

Checklist:

• Footer links to legal pages or org-approved legal destinations.
• Visible organization name and contact path.
• Company address or contact email where appropriate.
• About/team/founder evidence on at least one trust page.
• Product claims tied back to named company or project.

Rules:

• If legal pages live on the parent brand domain, keep that policy consistent and obvious.
• Do not ship anonymous company/product sites with no ownership signals.

---

P2 — Recommended On Most Sites

9. Every important page should have a dedicated citation target

Why it matters:

• AI systems and technical buyers cite specific pages, not whole sites.
• The best citation targets are comparison pages, product pages, architecture pages, benchmarks, and FAQ-rich explainers.

Checklist:

• Have at least one primary citable page per product or product line.
• Put strong factual claims, specs, comparison tables, and supporting evidence on that page.
• Make sure the URL is stable and linked from the homepage or docs hub.

Rules:

• Do not bury your strongest differentiator in a PDF, a removed page, or a hidden doc.

---

10. Every site should include an AI-readable brand and product summary

Why it matters:

• llms.txt and llms-full.txt are useful when they are concise, factual, and internally consistent.

Checklist:

• State what the company is.
• State what each product is.
• Link to the best docs/product URLs.
• Include trust facts: legal entity, grant, standards, notable integrations, certifications, or ecosystem standing.
• Add disambiguation where product names are ambiguous.

Rules:

• Do not let llms.txt contradict the main site copy or robots rules.

---

What Every New Page Should Pass Before Merge

Use this as the page-level gate:

• Unique title
• Unique meta description
• Canonical URL set to production
• One clear H1 with real topic/product keywords
• Correct OG/Twitter tags
• At least one relevant JSON-LD block
• Main content visible in initial HTML
• No dead links
• Linked from at least one existing page
• No unsupported technical claims

---

What Every New Site Or Major Relaunch Should Pass Before Launch

• robots.txt present and intentional
• sitemap.xml present and aligned with canonicals
• llms.txt present if AI discoverability is desired
• Production build only, no dev artifacts
• Preview environments blocked or noindex
• Security headers configured at the real serving layer
• Legal/trust footer in place
• Homepage, docs hub, and product pages all have distinct metadata and schema

---

Working Rule For The Org

From now on, no page should merge unless it is:

• identifiable
• canonicalized
• machine-readable
• crawlable
• trustable
• internally linked
• free of unsupported claims

That is the baseline standard for SEO across all org websites.