ensorahealth.com — AI Site Grade

Analysis

Ensora Health — AI-Visibility Audit

The llms.txt file, which is meant to be the authoritative AI-friendly content map, points every link at a staging subdomain (stg-ensora-staging.kinsta.cloud) rather than the live production domain — meaning any LLM consuming this file gets broken or wrong URLs.

Crawler Access

All major AI crawlers — GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, ChatGPT-User, Google-Extended, Applebot-Extended — receive a full 200 response with identical byte content (109,658 bytes) to a browser baseline. No UA-based blocking, no Cloudflare challenge, no JS shell. The robots.txt is unusually sophisticated: it explicitly allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, and Google-Extended with only admin-area exclusions, and even includes commentary explaining each bot's purpose. Bytespider is the only AI crawler disallowed at root. The site runs on Cloudflare behind Kinsta (WordPress hosting), with s-maxage=86400 cache headers that serve stale content up to a day.

Cold-Knowledge Gap

A frontier LLM queried cold describes Ensora Health as "an AI-driven clinical decision support company for hospitals and health systems" — a description that is entirely wrong. The actual Ensora Health is an EHR and practice management software vendor for mental health therapists (TheraNest) and physical/occupational/speech therapists (Fusion). The cold model hallucinated a hospital-facing AI analytics product. The site itself claims 200,000+ providers and 28,000+ practices served, yet the model has zero awareness of this scale or positioning.

Schema Posture

The homepage carries a rich WebPage/Organization/WebSite JSON-LD graph with logo, social profiles, breadcrumbs, and search action. Notably, it also includes an AggregateRating with ratingValue: 4.3 and reviewCount: 165 — a strong social-proof signal for AI answer surfaces. However, the product pages (TheraNest, Fusion) use only generic WebPage schema with no SoftwareApplication or Product markup, which limits how AI engines can classify and compare them against competitors. The comparison pages (e.g., TheraNest vs SimplePractice) do use FAQPage schema with structured Q&A — a positive signal.

External Signals

External search results for "Ensora Health" return zero organic results on DuckDuckGo — no press coverage, no review sites, no Reddit threads, no industry publications. The DNS TXT records reveal domain verification tokens for OpenAI, Anthropic, Google, Atlassian, Zoom, Calendly, Figma, Pendo, Zapier, and Lucid — indicating active integration and verification across the AI ecosystem. The G2 profile is referenced in schema (g2.com/sellers/ensora-health-formerly-therapy-brands) but does not surface in search results.

Surprising Details

The llms.txt (generated by Yoast SEO v27.6) contains 17 staging-domain URLs — every single page, post, case study, and pricing link points to stg-ensora-staging.kinsta.cloud instead of ensorahealth.com. This is a critical AI-visibility defect: any LLM consuming this file will attempt to cite or link to a staging environment that may be access-restricted or stale. The blog includes a post titled "How to Rank in ChatGPT and AI Tools" (published Feb 2026) — the site is actively producing AI-visibility content for its customers while having a broken llms.txt and a massive cold-knowledge gap about its own brand.

Findings

1. llms.txt points all links to staging subdomain High

   The llms.txt file contains 17 URLs pointing to stg-ensora-staging.kinsta.cloud instead of the live ensorahealth.com domain, making every link broken or wrong for LLMs consuming the file.

   What to change: Update the llms.txt file to replace all staging subdomain URLs with the corresponding production URLs on ensorahealth.com.

2. Frontier LLM hallucinates Ensora Health as hospital AI analytics vendor High

   A cold query to a frontier LLM describes Ensora Health as an AI-driven clinical decision support company for hospitals, which is entirely incorrect. The actual company provides EHR and practice management software for mental health and therapy practices, serving 200,000+ providers.

   What to change: Increase authoritative content about the company's actual offerings, such as detailed product pages, case studies, and press releases, to correct the AI knowledge gap.

3. Zero organic search results for brand name High

   Web searches for 'Ensora Health' return no organic results on DuckDuckGo, indicating no press coverage, review sites, or industry publications indexed. The G2 profile referenced in schema does not surface in search results.

   What to change: Build external signals through press releases, guest posts, directory listings, and active engagement on review platforms like G2 and Capterra.

4. Product pages lack SoftwareApplication schema Medium

   The TheraNest and Fusion product pages use only generic WebPage schema instead of SoftwareApplication or Product markup, limiting how AI engines can classify and compare these products against competitors.

   What to change: Add SoftwareApplication schema with properties like applicationCategory, operatingSystem, and offers to product pages.

5. Cache headers serve stale content up to a day Low

   The site sets s-maxage=86400 cache headers, meaning AI crawlers may receive content that is up to 24 hours old, which can be problematic for time-sensitive pages like pricing or blog posts.

   What to change: Reduce s-maxage to a shorter duration (e.g., 3600) for dynamic pages, or use cache-busting techniques for critical updates.

What's working

• Robots.txt explicitly allows major AI crawlers with commentary — The robots.txt file explicitly allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, and Google-Extended, with only admin-area exclusions, and includes explanatory comments for each bot's purpose.
• Homepage includes AggregateRating schema with 4.3 stars and 165 reviews — The homepage JSON-LD includes an AggregateRating with ratingValue 4.3 and reviewCount 165, providing strong social proof that can be surfaced in AI answer snippets.
• Comparison pages use FAQPage schema with structured Q&A — The comparison page (e.g., TheraNest vs SimplePractice) implements FAQPage schema with well-structured questions and answers, which helps AI engines extract comparative information.
• Blog post on ranking in ChatGPT and AI tools published — The site has published a blog post titled 'How to Rank in ChatGPT & AI Tools' (Feb 2026), demonstrating active content creation aimed at AI visibility.
• DNS TXT records show verification tokens for multiple AI platforms — The DNS TXT records include domain verification tokens for OpenAI, Anthropic, Google, Atlassian, Zoom, Calendly, Figma, Pendo, Zapier, and Lucid, indicating active integration and verification across the AI ecosystem.
• All major AI crawlers receive full 200 response with identical content — All tested AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) receive a full 200 response with the same byte content as a browser, with no UA-based blocking or JavaScript shells.