aurorasolar.com — AI Site Grade

Analysis

The llms.txt exists but points entirely to a staging subdomain that blocks all crawlers

Every single URL in Aurora Solar's /llms.txt uses staging.aurorasolar.com — a subdomain whose robots.txt issues Disallow: / to all user-agents. An AI crawler following the llms.txt would hit a dead end on every link. This is the most consequential AI-visibility gap on the site: a well-intentioned but self-defeating llms.txt.

Crawler Access

All major AI bots (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, OAI-SearchBot, Bytespider, Applebot-Extended) receive 200 with full content from the homepage — identical byte size (230,408) to browser baseline. No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt uses a Yoast-generated catch-all Disallow: (allow all) with no AI-specific rules. The site runs on Cloudflare behind Kinsta hosting (WordPress). The technical foundation for AI crawling is clean.

llms.txt: Present but Poisoned

/llms.txt returns 200 with 126KB of content — a rare and ambitious file. However, every single link points to staging.aurorasolar.com, a subdomain whose robots.txt blocks all crawlers with Disallow: /. The staging pages themselves return noindex, nofollow and link back to the staging subdomain internally. An AI crawler consuming this llms.txt would follow links to a walled-off staging environment, not the live site. The file also lacks a summary paragraph (only a bare > blockquote) and contains no structured sections for "For LLMs" vs "For Humans" — the emerging llms.txt convention.

Cold-Knowledge Gap

The LLM knows Aurora Solar as a $4B-valued, $250M+ funded solar design software company founded in 2013 by Christopher Hopper and Samuel Adeyemo, with mentions of 2023 layoffs (~20%) and mixed G2 reviews about pricing. The live site, however, makes no mention of valuation, funding amounts, or layoffs. The homepage leads with "The World's #1 Solar Design Software" and "90% of top residential US solar companies use Aurora" — claims the cold model cannot verify. The site's About page details the founding story and funding rounds (Series A through D) but omits the $4B valuation figure that dominates the model's prior. The gap: the model knows the company as a well-funded unicorn with recent turbulence; the site presents a polished, risk-free market leader narrative with no acknowledgment of industry headwinds.

Schema Posture

Every page carries the same three JSON-LD blocks: WebPage, BreadcrumbList, WebSite, and Organization — all correctly structured with @id references. However, no product schema (SoftwareApplication, Product) exists on any product page (Sales Mode, Design Mode, Aurora AI). No FAQPage schema despite the blog having FAQ content. No Review schema for the extensive testimonial blocks. The Organization schema has a logo but no sameAs links to social profiles, no foundingDate, no foundingLocation, and no numberOfEmployees. The WebSite schema includes SearchAction but the urlTemplate points to ?s={search_term_string} which is a WordPress-native search, not a dedicated site search endpoint.

Content Signals

The homepage is text-rich (3,200+ words) with strong comparison language ("We've used Solo, OpenSolar, and Solargraf...") and customer testimonials naming competitors directly. The blog has 46 pages of pagination with posts dating back to 2018 — significant content tail but much of it stale (COVID-19 stimulus posts, 2019 policy roundups). The sitemap indexes 576+ URLs across 9 sub-sitemaps. The site uses max-image-preview:large and max-snippet:-1 in robots meta, which is generous for AI snippet extraction.

External Signals

DNS records reveal a dense SaaS tooling stack: HubSpot, Drift, Segment, Pendo, Stripe, Atlassian, Notion, Miro, Rippling, Zoom, and Autodesk verifications. This signals a mature, well-resourced engineering org — making the staging-URL llms.txt oversight more surprising. No recent press or Reddit threads surfaced in search, suggesting the brand's external AI footprint is shaped primarily by its own content and funding announcements rather than ongoing community discussion.

Findings

1. llms.txt points entirely to staging subdomain that blocks all crawlers High

   Every URL in /llms.txt uses staging.aurorasolar.com, whose robots.txt disallows all crawlers. AI agents following the file hit a dead end.

   What to change: Replace all staging.aurorasolar.com URLs in /llms.txt with the corresponding live aurorasolar.com URLs.

2. No product schema on product pages Medium

   Product pages (Sales Mode, Design Mode, Aurora AI) lack SoftwareApplication or Product schema, reducing AI understanding of the software offering.

   What to change: Add SoftwareApplication schema to each product page with name, description, applicationCategory, and offers.

3. Organization schema missing sameAs, foundingDate, and employee count Medium

   The Organization JSON-LD lacks sameAs links to social profiles, foundingDate, foundingLocation, and numberOfEmployees, limiting entity enrichment.

   What to change: Add sameAs, foundingDate, foundingLocation, and numberOfEmployees to the Organization schema.

4. Site omits valuation and layoffs that dominate LLM prior knowledge Medium

   The LLM knows Aurora Solar as a $4B-valued company with 2023 layoffs, but the site makes no mention of valuation, funding amounts, or layoffs, creating a mismatch.

   What to change: Add a page or section that transparently addresses the company's funding, valuation, and recent history to align with external knowledge.

5. llms.txt lacks summary paragraph and structured sections Low

   The llms.txt contains only a bare blockquote and no summary paragraph or sections for 'For LLMs' vs 'For Humans', missing the emerging convention.

   What to change: Add a summary paragraph at the top and organize links under 'For LLMs' and 'For Humans' sections.

6. Blog contains stale content from 2019-2020 Low

   The blog has posts dating back to 2018, including COVID-19 stimulus posts and 2019 policy roundups, which may mislead AI models about current relevance.

   What to change: Review and update or remove outdated blog posts, especially those referencing specific years or events.

7. No FAQPage schema on blog or FAQ content Low

   Despite FAQ-style content on the blog, no FAQPage schema is used, missing an opportunity for rich results.

   What to change: Add FAQPage schema to pages with question-and-answer content.

8. No Review schema for testimonials Low

   Testimonial blocks on the homepage lack Review schema, reducing their visibility in AI-generated summaries.

   What to change: Add Review schema to customer testimonial sections.

What's working

• All major AI bots receive full content with no blocking — Homepage returns 200 with full content to all tested AI bots, with no UA-based blocking or Cloudflare challenges.
• llms.txt file exists and is large (126KB) — The site serves a 126KB llms.txt, a rare and ambitious file that signals intent to support AI crawlers.
• Generous robots meta directives for snippets and images — The site uses max-image-preview:large and max-snippet:-1, allowing AI models to extract rich snippets and large images.
• Homepage is text-rich with 3,200+ words — The homepage contains over 3,200 words of descriptive content, providing ample material for AI understanding.
• JSON-LD schemas are correctly structured with @id references — Every page includes WebPage, BreadcrumbList, WebSite, and Organization schemas with proper @id linking.
• Sitemap indexes 576+ URLs across 9 sub-sitemaps — The sitemap provides broad coverage of the site's content, aiding discovery by crawlers.
• Mature SaaS tooling stack signals well-resourced engineering — DNS records show integrations with HubSpot, Drift, Segment, Pendo, Stripe, and others, indicating a sophisticated tech operation.