saatva.com — AI Site Grade

Analysis

I have enough data to write a sharp audit. Let me compile the findings.

Saatva.com — AI-Visibility Audit

The site's most consequential AI-visibility gap is not blocked crawlers or missing schema — it is the complete absence of an llms.txt file (404), which means AI engines have no structured, permissioned content map to reference, forcing them to rely on their own crawl prioritization across 817+ sitemap URLs.

Crawler Access

Every major AI crawler — GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Applebot-Extended, Bytespider, anthropic-ai — receives a full 200 response with identical byte payload (984,828 bytes) as a browser. No UA-based blocking exists. The site runs on AWS CloudFront behind a Next.js stack with HSTS and X-Frame-Options: SAMEORIGIN. The robots.txt uses a catch-all * rule with no AI-bot-specific directives, disallowing only search-parameter URLs and blog tag/search paths. This is permissive but undifferentiated — Saatva treats all bots identically, missing the opportunity to grant premium access paths to AI crawlers.

Cold-Knowledge Gap

The LLM's cold knowledge accurately describes Saatva as a DTC luxury mattress company with white-glove delivery, a 365-night trial, and the Saatva Classic flagship. However, the model does not know about Saatva's expansion into furniture, bedding, rugs, lighting, and candles — a full home-goods ecosystem that dominates the homepage's heading structure. The model also misses the Team USA / LA28 Olympic partnership (Official Mattress and Restorative Sleep Provider), which is a major brand signal the site prominently features on the about page, press room, and blog. The model's knowledge of the Loom & Leaf sub-brand is stale — the site now routes /mattresses/loom-and-leaf but the brand has been folded into the Contour5 / Memory Foam Hybrid lineup.

Schema Posture

Product pages carry strong schema: Product, AggregateOffer, AggregateRating (4.8 from 5,671 reviews), MerchantReturnPolicy (365-day free return), WarrantyPromise (P100Y), and FAQPage with detailed Q&A. The homepage, however, carries only a single Offer schema for the sitewide sale — no Organization, WebSite, or Product markup at the root level. The about-us page has no Organization schema despite being the canonical brand-identity page. The blog and press room are WordPress subdirectories with their own schema graphs (Person, WebSite, BlogPosting, VideoObject) but these are disconnected from the main site's schema namespace.

Content & Answer Signals

The homepage is a dense product catalog with 2,100+ words of visible text, strong comparison language ("Compare our mattresses"), and FAQ patterns. The compare page (/mattresses/compare) offers a side-by-side interactive tool but renders as a JavaScript-dependent widget — the plain GET returns only the shell with dropdown labels and no actual comparison data in the HTML. AI crawlers that do not execute JS will see an empty comparison interface. The blog has rich, structured content (buying guides, sleep health, athlete features) but sits on a separate WordPress subdirectory with its own sitemap (/blog/sitemap.xml).

External Signals

Saatva links to BBB, Trustpilot, ResellerRatings, and Google Store reviews from the reviews page. The press room documents recent awards: Great Place to Work 2026, PureWow Readers' Favorite, Newsweek Readers' Choice Best Mattress 2026, and CEO Ron Rudzin named EY Entrepreneur Of The Year 2026 finalist. The Team USA partnership is the most externally citable brand signal, yet it is absent from the LLM's cold knowledge — a gap that directly impacts how AI-generated answers describe the brand's authority and reach.

Findings

1. No llms.txt file published High

   The site returns a 404 for /llms.txt, meaning AI engines have no structured, permissioned content map to reference, forcing reliance on crawl prioritization across 817+ sitemap URLs.

   What to change: Publish an llms.txt file that lists key content paths (product pages, about, press room, blog) and optionally an llms-full.txt for deeper content.

2. LLM cold knowledge misses home-goods ecosystem Medium

   The LLM's cold knowledge describes Saatva as a mattress company but does not know about its expansion into furniture, bedding, rugs, lighting, and candles, which dominate the homepage heading structure.

   What to change: Add structured data (e.g., ItemList, Product) for each product category on the homepage and ensure the site is well-represented in external knowledge sources.

3. LLM cold knowledge misses Team USA partnership Medium

   The LLM does not know about Saatva's role as Official Mattress and Restorative Sleep Provider for Team USA / LA28 Olympics, a major brand signal prominently featured on the about page, press room, and blog.

   What to change: Publish a press release or news article about the partnership with structured data (NewsArticle, Organization) and ensure it is indexed and linked from the homepage.

4. Comparison page renders as empty JS shell High

   The /mattresses/compare page is a JavaScript-dependent widget; the plain GET returns only the shell with dropdown labels and no actual comparison data in the HTML, so AI crawlers that do not execute JS see an empty interface.

   What to change: Server-side render the comparison data or include a static HTML fallback with the key product differences.

5. Homepage lacks Organization and WebSite schema Medium

   The homepage carries only a single Offer schema for the sitewide sale, with no Organization, WebSite, or Product markup at the root level, reducing its ability to convey brand identity to AI engines.

   What to change: Add Organization, WebSite, and Product (for the flagship mattress) schema to the homepage.

6. About-us page has no Organization schema Medium

   The about-us page, the canonical brand-identity page, lacks Organization schema, missing an opportunity to reinforce brand signals for AI knowledge graphs.

   What to change: Add Organization schema with name, logo, description, and sameAs links to the about-us page.

7. All AI crawlers treated identically with no premium paths Low

   The robots.txt uses a catch-all * rule with no AI-bot-specific directives, missing the opportunity to grant premium access paths (e.g., to the blog or press room) to AI crawlers.

   What to change: Add specific directives for AI crawlers (e.g., Allow: /blog, Allow: /press-room) to prioritize important content.

8. Blog and press room schema disconnected from main site Low

   The blog and press room are WordPress subdirectories with their own schema graphs (Person, WebSite, BlogPosting, VideoObject) but these are disconnected from the main site's schema namespace, reducing coherence.

   What to change: Ensure the blog and press room schema reference the main Organization schema via @id or sameAs.

9. Loom & Leaf sub-brand knowledge is stale Low

   The LLM's cold knowledge references Loom & Leaf, but the site now routes /mattresses/loom-and-leaf and the brand has been folded into the Contour5 / Memory Foam Hybrid lineup, causing potential confusion.

   What to change: Update the Loom & Leaf page with clear redirects or schema indicating the new product name, and ensure external knowledge sources are updated.

What's working

• All major AI crawlers fully allowed — Every major AI crawler receives a full 200 response with identical content as a browser, with no UA-based blocking.
• Product pages carry rich structured data — Product pages include Product, AggregateOffer, AggregateRating (4.8 from 5,671 reviews), MerchantReturnPolicy, WarrantyPromise, and FAQPage schema.
• Blog has rich, structured content — The blog offers buying guides, sleep health articles, and athlete features with its own schema (BlogPosting, VideoObject), providing valuable content for AI engines.
• Press room documents recent awards and partnerships — The press room lists awards (Great Place to Work 2026, Newsweek Readers' Choice Best Mattress 2026) and the Team USA partnership, providing citable brand signals.
• Reviews page links to third-party review platforms — The reviews page links to BBB, Trustpilot, ResellerRatings, and Google Store reviews, building external credibility signals.
• Homepage uses FAQ patterns and comparison language — The homepage includes strong comparison language and FAQ patterns, which can help AI engines extract answer snippets.