civitai__civitai/docs/seo-audit.md

SEO Audit — Base Document

A running reference for auditing Civitai's SEO posture. This is the "what we know and where the gaps are" doc — findings and fixes should be appended as sections below.

Started: 2026-04-24 (Briant).

---

Context: .com vs .red

Civitai runs under two canonical hosts driven by the same codebase:

Domain	Role	Content
civitai.com (green)	SFW site	Content whose nsfwLevel passes hasSafeBrowsingLevel
civitai.red (red / blue)	NSFW site	Everything, including mature content

The domain is resolved server-side in _app.tsx:289 (canIndex = serverDomainMap values includes host) and surfaced to React via useDomainColor() in src/hooks/useDomainColor.tsx. Feature-flag style: isGreen = .com, isBlue / isRed = .red variants.

Crawler behavior at the domain boundary

When a user requests an NSFW resource on .com:

1. SSR renders the page.
2. <Meta> is rendered outside <SensitiveShield> so meta tags are always present.
3. Inside the shield, a replacement card ("Mature content has a new home → civitai.red") is rendered instead of the real body.
4. Googlebot (logged-out, safe-level only) receives: canonical title + description for the article, but a body that redirects off-site.

This creates a content/meta mismatch if the URL remains indexable on .com. The fix pattern is to toggle deIndex based on both nsfwLevel and the domain: when we're on green and content is NSFW, emit noindex,nofollow so the .red URL wins the canonical fight.

Meta component safeguards

Meta already does the right thing for images: line 75 only selects getIsSafeBrowsingLevel images for og:image. The primary source is the /api/og endpoint, which is expected to render a SFW preview card regardless of content rating. If we ever allow raw-image fallbacks for NSFW content, this guard is the last line of defense.

deIndex, canonical, and alternate are page-level controls. Each page is responsible for passing the right values.

Global indexing gate

canIndex is true whenever the host matches a canonical server domain. It is NOT sensitive to content rating. .red is currently in canIndex: true territory; whether we want .red content indexed at all is a product question — see Open Questions below.

---

Checklist: what a detail-page audit should cover

For every detail page (models, images, posts, articles, collections, user profiles, bounty, club, etc.):

• <Meta> is outside any visibility gate (SensitiveShield, login gate, mature-content blur).
• description is derived from content that is safe to show in a SERP snippet, or the page is de-indexed when the content is NSFW.
• og:image uses either /api/og or a SFW-filtered selector (never an unfiltered image.url).
• deIndex is set when any of: - content is unpublished / draft, - availability === Unsearchable, - we're on .com and the content is NSFW.
• canonical points at the production URL with slug (or the slug-less URL if that's the canonical choice — be consistent).
• alternate is set if slug-less and slug'd URLs both resolve.
• JSON-LD schema is present where it adds value (Article, Product, Person, ImageObject, etc.) — optional, but worth noting.

---

Audit checklist

Tick a box once the page has been audited against the checklist above and an entry has been added to Findings below (even if no fix was needed — a "no issues" note counts).

P0 — Detail pages with NSFW content

Top priority: these are the pages most likely to produce the content/meta mismatch described above. All have user-rated content; all need the NSFW-aware deIndex guard.

• /articles/:id/:slug? — articles/[id]/[[...slug]].tsx (fixed 2026-04-24)
• /models/:id/:slug? — models/[id]/[[...slug]].tsx
• /model-versions/:id — model-versions/[id].tsx
• /images/:imageId — images/[imageId].tsx
• /posts/:postId/:postSlug? — posts/[postId]/[[...postSlug]].tsx
• /bounties/:id/:slug? — bounties/[id]/[[...slug]].tsx
• /bounties/:id/entries/:entryId — bounties/[id]/entries/[entryId]/index.tsx
• /collections/:collectionId — collections/[collectionId]/index.tsx
• /comics/:id/:slug? — comics/[id]/[[...slug]].tsx

P1 — User-facing pages with mixed or derived content ratings

These aggregate user content (so they can surface NSFW previews in meta) and are SEO-visible.

• /user/:username — user/[username]/index.tsx
• /user/:username/models — user/[username]/models.tsx
• /user/:username/images — user/[username]/images.tsx
• /user/:username/posts — user/[username]/posts.tsx
• /user/:username/videos — user/[username]/videos.tsx
• /user/:username/articles — user/[username]/articles.tsx
• /user/:username/collections — user/[username]/collections.tsx
• /user/:username/comics — user/[username]/comics.tsx
• /user/:username/:list — user/[username]/[list].tsx (catch-all list route)
• /user-id/:userId — user-id/[userId].tsx (confirm canonical points at username URL)
• /tag/:tagname — tag/[tagname].tsx (tag aggregation, can include NSFW)
• /reviews/:reviewId — reviews/[reviewId].tsx
• /comments/v2/:id — comments/v2/[id].tsx (likely should de-index)
• /challenges/:id/:slug? — challenges/[id]/[[...slug]].tsx
• /events/:slug — events/[slug].tsx
• /leaderboard/:id — leaderboard/[id].tsx
• /auctions/:slug? — auctions/[[...slug]].tsx
• /tools/:slug — tools/[slug].tsx

P2 — Index / feed pages

High traffic, but meta is mostly static per route. Confirm titles, descriptions, and canonicals are set (not inherited from defaults).

• / — index.tsx
• /home — home/index.tsx
• /models — models/index.tsx
• /images — images/index.tsx
• /videos — videos/index.tsx
• /posts — posts/index.tsx
• /articles — articles/index.tsx
• /bounties — bounties/index.tsx
• /collections — collections/index.tsx
• /comics — comics/index.tsx
• /comics/browse — comics/browse.tsx
• /challenges — challenges/index.tsx
• /challenges/winners — challenges/winners.tsx
• /events — events/index.tsx
• /tools — tools/index.tsx
• /builds — builds/index.tsx
• /search/models — search/models.tsx
• /search/images — search/images.tsx
• /search/articles — search/articles.tsx
• /search/bounties — search/bounties.tsx
• /search/collections — search/collections.tsx
• /search/comics — search/comics.tsx
• /search/tools — search/tools.tsx
• /search/users — search/users.tsx

P3 — Marketing / evergreen / legal

Static (or near-static) pages that should be fully indexed on .com and may benefit from richer meta/schema.

• /content/:slug* — content/[[...slug]].tsx (TOS/Privacy/etc. catch-all)
• /pricing — pricing/index.tsx
• /safety — safety/index.tsx
• /newsroom — newsroom/index.tsx
• /support — support/index.tsx
• /creator-program — creator-program/index.tsx
• /changelog — changelog/index.tsx
• /product/vault — product/vault.tsx
• /product/link — product/link.tsx
• /product/odor — product/odor.tsx
• /shop — shop/index.tsx
• /gift-cards — gift-cards/index.tsx
• /buzz/marketplace — buzz/marketplace/index.tsx

P4 — Sitemaps

Must agree with per-page deIndex decisions (submitting a noindex URL is a contradictory signal).

• /sitemap-models.xml — sitemap-models.xml/index.tsx
• /sitemap-articles.xml — sitemap-articles.xml/index.tsx
• /sitemap-tools.xml — sitemap-tools.xml/index.tsx
• Decide whether to add missing sitemaps: posts, images, bounties, collections, comics, users, events, challenges.

Should be de-indexed (sanity-check only)

These don't need SEO polish but do need to reliably send noindex so they don't clutter SERPs. Spot-check one page per group; if all members of the group render noindex, tick the group.

• Edit / create / wizard routes (articles/[id]/edit, articles/create, models/[id]/edit, models/[id]/wizard, models/[id]/model-versions/…, models/create, models/train, posts/[postId]/edit, posts/create, bounties/[id]/edit, bounties/create, bounties/[id]/entries/create, bounties/[id]/entries/[entryId]/edit, comics/create, comics/project/[id]/*, train, training/[workflowId], generate)
• Account / billing routes (user/account, user/notifications, user/membership, user/buzz-dashboard, user/downloads, user/transactions, user/vault, user/referrals, user/earn-potential, user/pool-estimate, user/stripe-connect/onboard)
• Auth / claim / redirect (login, login/token, verify-email, redirect, region-blocked, preview-restricted, redeem-code, claim/buzz/[id], claim/cosmetic/[id], intent/avatar, intent/post, discord/link-role, studio/confirm, subscribe/[plan], purchase/buzz)
• Payments (payment/*, tipalti/setup)
• Collections helpers (collections/[collectionId]/join, collections/[collectionId]/review, collections/youtube/auth)
• Games / one-offs (games/chopped, games/knights-of-new-order, dev/onboarding, data-graph-v2, images/iterate)
• Moderator / internal (moderator/*, research/*, testing/*)

---

Findings

Articles — /articles/:id/:slug?

File: src/pages/articles/[id]/[[...slug]].tsx

Status: Fixed 2026-04-24.

• ✅ <Meta> outside <SensitiveShield>.
• ✅ og:image uses /api/og?type=article&id=…; image fallback is SFW-filtered by Meta.
• ✅ deIndex now also fires when domain === 'green' and !hasSafeBrowsingLevel(article.nsfwLevel).
• ⚠️ description is truncate(removeTags(article.content), 150) — can leak explicit text into SERPs if an NSFW article ever slips through to the indexable set. Mitigated in practice by the green-domain deIndex guard, but still a risk for .red if we decide to index it.

---

Open Questions

1. Should .red be indexable at all? Currently canIndex is true on .red based purely on host match. Indexing NSFW URLs is legitimate (there's demand for mature search), but it has implications for ad networks, safe-search compliance, and backlink profile. Decide this explicitly rather than letting it be a side effect of the canIndex = host check.
2. Canonical format consistency. Articles use /articles/:id/:slug as canonical with /articles/:id as alternate. Other entity types should follow the same convention or we should document the divergence.
3. /api/og resilience. The OG endpoint is the primary social image for all shared links. Confirm it always renders a SFW card (including for NSFW-but-blurred-preview cases) and that it fails gracefully (we serve something, not a 500, if the entity is missing).
4. Sitemap coverage. Do our sitemaps respect the same green/NSFW de-index rules as the pages themselves? A sitemapped URL that returns noindex is a contradictory signal to Google.

---

Tooling notes

• Google Search Console is the source of truth for indexed-vs-submitted counts and coverage errors; pair any code finding with a GSC lookup to confirm real impact.
• For per-page verification, curl -A "Googlebot" <url> against the deployed site is the fastest way to see exactly what the crawler receives (meta tags and body copy).
• Dev-server rendering is sufficient for checking meta tags but not for verifying full SSR output — use a production build (pnpm run build + pnpm run start) when the server-render path matters.