AI Site Grade
better.com — AI Site Grade
Better.com's SPA architecture and sanitized narrative create a severe AI visibility gap, with cold LLM knowledge dominated by the 2021 layoff controversy.
Better.com's Next.js SPA delivers a 175-word shell to AI crawlers, its FAQ schema is empty, and cold LLM knowledge is dominated by the 2021 layoff controversy, creating a severe gap between the brand's sanitized narrative and AI perception.
- Findings
- 12
- Evidence checks
- 25
- Completed
- 30 May 2026
Analysis
I have enough data to write a sharp audit. Let me compile.
Better.com: AI crawlers see a JS shell, not a mortgage lender
Every AI crawler that hits better.com gets a 200 status and a ~57KB HTML page — but the homepage yields only 175 words of visible text, almost entirely navigation labels and a "Loading" placeholder. The site is a Next.js SPA served behind CloudFront, and while no bot is blocked at the HTTP level, the content AI engines actually ingest is a thin navigation skeleton, not the mortgage product the brand sells.
Crawler Access
The robots.txt is a bare-minimum file (112 bytes) with a single Allow: / rule for User-agent: * and no AI-bot-specific directives whatsoever. GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and all other tested bots receive identical 200 responses with identical byte sizes (~57KB) as a browser — no UA-based blocking. However, the llms.txt returns a 404 (serving the site's 404 page HTML), meaning there is no AI-friendly content map. The site runs on AWS CloudFront with a strict Content-Security-Policy and HSTS preload, but the SPA architecture means bots that do not execute JavaScript see only the app shell.
Cold-Knowledge Gap
The cold LLM knowledge about Better.com is dominated by the 2021 Zoom-call layoff of 900 employees and the failed SPAC merger — the single most prominent fact the model recalls. The site itself contains zero mention of these events. The about-us page presents a polished timeline (2014–Today) that skips from 2019 partnerships straight to 2022's "$100B funded" milestone, omitting 2020–2021 entirely. The model knows about the CEO Vishal Garg controversy; the site links to "Read Vishal's story" but the linked content pages (/content/vishal-garg-better-ceo, /content/the-house-that-got-away-why-i-started-better-mortgage) were not fetched but are listed in the internal links. The gap between the model's reputational baggage and the site's sanitized narrative is severe.
Schema Posture
Every page carries the same FinancialService JSON-LD block — identical name, description, address, telephone, and openingHours (24/7). The FAQ page (/faq) declares a FAQPage schema but with an empty mainEntity array — no actual Q&A pairs are marked up, defeating the purpose of FAQ schema entirely. The mortgage-rates page uses a MortgageLoan schema with hardcoded example values (2.5% interest, $400K amount) that do not reflect dynamic current rates. No Product, Review, AggregateRating, or HowTo schemas are used anywhere despite the site having a reviews page and educational content.
Content & Answer Signals
The homepage has zero H1–H3 headings in the extracted HTML — all headings are client-side rendered. The /content blog has 1,189+ URLs in the sitemap and publishes daily mortgage rate updates, but the blog post titled "Why does chatGPT say Better Mortgage is good for first-time buyers?" (visible in the /content listing) returns a 404 when accessed directly. The crypto-backed mortgages page is the most content-rich page found (1,844 words) with a comparison table (traditional vs. token-backed approach), FAQ sections, and a lead-capture form — but it lacks any structured data for the comparison or FAQ content.
Findings
Homepage delivers only 175 words of visible text to AI crawlers High
The homepage is a Next.js SPA that returns a 200 status but only 175 words of visible text, mostly navigation labels and a loading placeholder. AI crawlers that do not execute JavaScript see an empty shell.
What to change: Implement server-side rendering or static generation for key pages to ensure meaningful content is available in the initial HTML response.
llms.txt returns 404, no AI-friendly content map Medium
The llms.txt endpoint returns a 404 page, meaning there is no AI-friendly content map for crawlers to discover key resources.
What to change: Create an llms.txt file listing the most important pages for AI crawlers, such as mortgage rates, FAQ, and educational content.
FAQ page has empty mainEntity array in JSON-LD High
The FAQ page declares a FAQPage schema but the mainEntity array is empty, providing no structured Q&A data to search engines or AI crawlers.
What to change: Populate the mainEntity array with actual Question/Answer pairs for each FAQ item on the page.
MortgageLoan schema uses hardcoded example values Medium
The mortgage-rates page includes a MortgageLoan schema with static example values (2.5% interest, $400K amount) that do not reflect current dynamic rates, which could mislead AI crawlers.
What to change: Update the MortgageLoan schema to reflect current rates dynamically or remove the hardcoded example.
Cold LLM knowledge dominated by 2021 layoff controversy High
The most prominent fact recalled by LLMs about Better.com is the 2021 Zoom-call layoff of 900 employees and the failed SPAC merger. The site's about-us page omits this period entirely, creating a gap between AI perception and the brand's narrative.
What to change: Publish a transparent page addressing the layoff and subsequent changes to rebuild trust and provide a balanced narrative for AI crawlers.
Blog post about ChatGPT returns 404 Medium
A blog post titled 'Why does chatGPT say Better Mortgage is good for first-time buyers?' listed in the /content directory returns a 404 when accessed directly, wasting crawl budget and providing a poor user experience.
What to change: Restore the page or remove the link from the content listing to avoid broken links.
Homepage has zero H1-H3 headings in initial HTML High
The homepage's initial HTML contains no H1-H3 headings; all headings are rendered client-side via JavaScript, making the page appear structureless to crawlers that do not execute JS.
What to change: Include at least one H1 heading in the server-rendered HTML to provide a clear content signal.
Reviews page lacks Review or AggregateRating schema Medium
The about-us/reviews page contains customer testimonials but no structured data markup (Review, AggregateRating) to help search engines and AI crawlers understand the review content.
What to change: Add Review and AggregateRating schemas to the reviews page to enable rich snippets and improve AI understanding.
Robots.txt has no AI-bot-specific directives Low
The robots.txt file is minimal with only a single Allow rule for all user agents. No specific directives for GPTBot, ClaudeBot, or other AI crawlers exist, leaving access control unoptimized.
What to change: Consider adding specific rules for AI crawlers if certain sections should be excluded or prioritized.
Homepage visible text is only 175 words High
The homepage's extracted text is only 175 words, mostly navigation and a loading placeholder, providing almost no substantive content for AI crawlers.
What to change: Ensure the homepage includes meaningful introductory text in the initial HTML, such as a value proposition and key features.
No Product or HowTo schema on educational pages Medium
Educational content pages like the crypto-backed mortgages page lack Product or HowTo schemas, missing opportunities to provide structured guidance to AI crawlers.
What to change: Add HowTo schema for step-by-step guides and Product schema for mortgage products to enhance AI understanding.
About-us timeline omits 2020-2021 period Medium
The about-us page presents a timeline that jumps from 2019 partnerships to 2022 milestones, skipping the 2020-2021 period when the layoff occurred, creating a gap that may appear evasive to AI crawlers.
What to change: Include the 2020-2021 period in the timeline with appropriate context to provide a complete historical narrative.
What's working
- No AI crawlers are blocked at the HTTP level — All tested AI crawlers (GPTBot, ClaudeBot, etc.) receive 200 responses with identical content, indicating no UA-based blocking.
- Robots.txt allows all crawlers — The robots.txt file has a single Allow: / rule for all user agents, ensuring no pages are inadvertently blocked.
- Sitemap is available and contains 80 URLs — A sitemap is served at /sitemap.xml with 80 URLs, helping crawlers discover key pages.
- FinancialService JSON-LD present on all pages — Every page includes a FinancialService schema with consistent name, description, address, and 24/7 opening hours, providing basic business information to AI crawlers.
- Crypto-backed mortgages page is content-rich with 1,844 words — The crypto-backed mortgages page contains substantial text, a comparison table, and FAQ sections, providing valuable educational content.
- Blog publishes daily mortgage rate updates — The /content blog has over 1,189 URLs in the sitemap and publishes daily mortgage rate updates, indicating a commitment to fresh content.
- Betsy AI assistant page provides clear product information — The Betsy page (467 words) describes the AI assistant's features and benefits, offering useful content for AI crawlers interested in the company's AI capabilities.
- Strong security headers including HSTS preload — The site uses AWS CloudFront with strict Content-Security-Policy and HSTS preload, ensuring secure connections and trust signals.
Track better.com across AI search
This is one snapshot. Open the interactive report to inspect evidence, or grade another site free.