We Audited Stripe, Webflow, Atlassian, Datadog, and Auth0 for AI Readability. Here's What We Found.
Stripe, Webflow, Atlassian, Datadog, and Auth0 spend millions on web infrastructure. Their AI readability scores tell a different story. We ran each homepage through our 27-signal analyzer and published the raw findings.
We built Brandioz to measure how AI crawlers read websites. So we did what seemed obvious: we ran it on some of the most well-resourced software companies in the world. Stripe. Webflow. Atlassian. Datadog. Auth0. Companies with dedicated SEO teams, world-class engineers, and millions in web infrastructure.
The results were surprising — not because the scores were low, but because the failures were identical across all five. The same problems, repeated at scale.
Companies with FAQPage schema on their homepage
The highest-impact schema type for AI citation — missing from every single site we audited
The methodology
We used Brandioz's analyze endpoint to fetch each homepage as an AI crawler would — raw HTML, no JavaScript execution. The tool measures 27 signals across two scores: an AI Content Score covering what crawlers read, and a Crawl Score covering whether they can find and access the site. Both scores run 0–100.
- AI Content Score — title clarity, meta description, hero word count, paragraph density, heading structure, original statistics, FAQ headings, author signals, freshness signals, OG completeness
- Crawl Score — schema placement, FAQPage presence, Organization schema, CSR severity, robots.txt AI crawler permissions, sitemap, llms.txt, Common Crawl indexing, canonical URL
Every finding below comes from raw HTML — exactly what GPTBot, PerplexityBot, and ClaudeBot receive when they crawl these sites. No browser rendering. No JavaScript. Just the initial document.
Stripe — 62.09 / 100
Stripe AI Content Score — Crawl Score: 76.6
Strong infrastructure, weaker content signals
Stripe is the strongest performer of the five. Server-rendered with 2,010 words visible in raw HTML. All AI crawlers explicitly allowed. llms.txt present and valid at 7,428 words. Common Crawl indexed. The crawl infrastructure is exemplary — 76.6 crawl score.
- Passing: 2,010 words in raw HTML, all AI crawlers allowed, llms.txt valid at 7,428 words, Common Crawl indexed, strong meta description
- Primary failure: Organization and WebSite schema in `<body>` only — not `<head>`. AI crawlers that don't execute JavaScript skip body-only schema entirely.
- Missing: FAQPage schema, freshness signal, `max-snippet` directive
- Hero problem: Only 27 words in the hero — below the 40-word threshold where AI extraction becomes reliable
Stripe's hero reads: 'Financial infrastructure to grow your revenue.' Clear category signal — but only 27 words before hero copy becomes navigation. AI front-weights hero content disproportionately.
Webflow — 69.68 / 100
Webflow AI Content Score — Crawl Score: 71.4
Highest content score of the five, but critical schema placement errors
Webflow scores highest on content signals — 3,761 words in `<main>`, 16 quantified statistics, 12 FAQ questions detected in body HTML. But all four schema types (Organization, SoftwareApplication, WebPage, WebSite) are in `<body>`, not `<head>`. A platform that markets itself as AI-optimized has its own schema in the wrong location.
- Passing: 3,761 words server-rendered, 16 statistics, FAQ content in HTML, llms.txt valid, all AI crawlers allowed
- Primary failure: 4 schema types in `<body>` only. FAQPage schema missing despite 12 FAQ questions present in the HTML.
- Critical issue: 2 H1 tags — 'Make your website a growth engine' and 'Make websites that drive results' both tagged as H1. AI dilutes the primary identity signal across both.
- Missing: FAQPage schema, freshness signal, `og:url` tag
Webflow has the FAQ content. It has the statistics. It has the words. But the schema is in the wrong place, and two H1s split the identity signal Google and AI both rely on.
Atlassian — 59.81 / 100
Atlassian AI Content Score — Crawl Score: 75.5
Schema correctly in <head>, but content extracted from <main> is thin
Atlassian gets schema placement right — Organization, SoftwareApplication, WebPage, and WebSite all in `<head>`. They also have both llms.txt (2,768 words) and llms-full.txt, the most complete AI discovery setup of the five. But only 382 words are extracted from `<main>` despite 2,570 visible on the page.
- Passing: Schema correctly in `<head>`, llms.txt and llms-full.txt present, all AI crawlers allowed, 1,000 URLs in sitemap
- Primary failure: Only 382 words in `<main>`. Most content lives outside the semantic container AI weights most heavily.
- Missing: FAQPage schema, canonical URL, all 5 OpenGraph tags (0/5), freshness signal
- 2 H1 tags: Same dilution problem as Webflow
Atlassian OpenGraph tags present
Complete absence — social shares and some AI tools lose all metadata context
Datadog — 42.67 / 100
Datadog AI Content Score — Crawl Score: 84.6
Highest crawl score of the five. Lowest content score. A paradox worth studying.
Datadog is the most interesting case in this audit. Their crawl infrastructure is the best of the five — schema correctly in `<head>`, freshness signal present, llms.txt at 57,330 words (by far the largest), all AI crawlers allowed. But the AI content score is 42.67 — lowest of the group. The reason: only 67 words extracted from the page.
- Passing: Schema in `<head>`, freshness signal present, llms.txt at 57,330 words, all crawlers allowed, canonical declared, OG tags complete at 100%
- Primary failure: Only 67 words extracted despite 1,638 visible on the page. The content lives outside semantic containers AI weights.
- Zero scores: Hero word count 0, paragraph density 0, heading count ratio 0, H1-meta alignment 0, statistics 0
- Meta description: Only 54 characters — well below the 120-character threshold where AI has enough context
Datadog's llms.txt is 57,330 words — the most thorough of any site we audited. Their robots.txt explicitly allows every major AI crawler. But the homepage itself gives AI crawlers 67 words to work with. Infrastructure without content is a partial solution.
Datadog llms.txt word count — largest of the five
Exceptional AI navigation infrastructure. The homepage content gap is the single fix that would move the score significantly.
Auth0 — 57.75 / 100
Auth0 AI Content Score — Crawl Score: 85.2
Highest crawl score alongside Datadog. Strong infrastructure, content gaps in hero and statistics.
Auth0 has the strongest schema setup of the five — BreadcrumbList, Organization, SoftwareApplication, WebPage, and WebSite all correctly in `<head>`. llms.txt and llms-full.txt both present. 2,674 words in `<main>`. 13 descriptive H2s. The fundamentals are right.
- Passing: Schema correctly in `<head>` with 5 types, llms.txt and llms-full.txt present, 2,674 words in `<main>`, 13 descriptive H2s, OG tags complete at 85%, all AI crawlers allowed
- Primary failure: FAQPage schema missing. Hero only 19 words. Zero statistics or quantified claims.
- Missing: FAQPage schema, freshness signal, `max-snippet` directive
- Irony: Auth0 has speakable schema implemented — a rare signal most sites miss — but is missing FAQPage schema, which has higher citation impact
Auth0 implemented speakable schema — which tells AI systems which parts of a page are most citable. That's an advanced GEO signal most large companies haven't touched. The gap is FAQPage schema and statistics, not the infrastructure.
The pattern across all five
Five companies. Five different tech stacks — Next.js/Vercel (Stripe, Auth0), Webflow (Webflow), custom (Atlassian), Hugo/Tailwind (Datadog). Five different budgets. The same failures repeating across all of them.
- FAQPage schema: 0 of 5 — Not one homepage has it. Every site has FAQ-like content somewhere. None have the schema that makes it machine-readable for AI citation.
- Freshness signal: 1 of 5 — Only Datadog has a `<time>` tag or article date meta. Retrieval-first platforms like Perplexity deprioritize undated content.
- Schema in wrong location: 2 of 5 — Stripe and Webflow have all schema in `<body>`. Atlassian, Datadog, and Auth0 correctly place it in `<head>`.
- Multiple H1 tags: 2 of 5 — Webflow and Atlassian both split their primary identity signal across two competing H1s.
- Zero statistics: 3 of 5 — Atlassian, Datadog, and Auth0 have no quantified claims in AI-readable HTML. Original data is the highest-leverage citation signal across all AI platforms.
- Thin hero content: 4 of 5 — Stripe (27 words), Webflow (150 words ✓), Atlassian (24 words), Datadog (0 words), Auth0 (19 words). Only Webflow clears the 40-word threshold.
Companies with hero sections below the 40-word AI extraction threshold
The section AI weights most heavily — thin on nearly every homepage we audited
What this means for your site
If Stripe, Webflow, Atlassian, Datadog, and Auth0 — with their combined engineering resources — haven't fixed these issues, the realistic probability that your competitors have is very low. These aren't obscure optimizations. They're the baseline requirements for AI crawler readability, and the bar is currently low across the entire SaaS industry.
- Add FAQPage schema to `<head>` — 5 questions covering what you are, who you serve, how you work, what it costs, and how you differ. This is the single fix missing from every site in this audit.
- Move schema to `<head>` if it's in `<body>` — run `curl https://yourdomain.com | grep -o 'application/ld+json'` and check whether those blocks appear before or after `</head>`.
- Fix multiple H1s — one H1 per page. AI treats every H1 as an equally-weighted identity claim.
- Add a `<time>` tag — any visible date gives retrieval-first platforms a freshness signal. Datadog is the only site here that has one.
- Expand your hero — `curl -A 'GPTBot' https://yourdomain.com` and count words in your hero section in the raw output. Under 40 means AI is front-weighting near-empty content.
- Add quantified claims — 'used by 10,000+ teams', '99.9% uptime', '5-minute integration'. Three of five sites audited have zero statistics in AI-readable HTML.
Run the same audit on your own site at brandioz.com/dashboard — free, no sign-in required. You'll see exactly which of these 27 signals you're passing and failing, with the specific HTML causing each issue.
Key takeaway
Even the most sophisticated SaaS companies have significant AI readability gaps. The most common failure across all five: FAQPage schema missing from every single homepage. Schema in the wrong location, no freshness signals, and thin hero content repeat across companies regardless of tech stack. These are fixable in hours. If Stripe, Webflow, Atlassian, Datadog, and Auth0 haven't fixed them — your competitors almost certainly haven't either.
See how your site scores
Free AI visibility analysis — takes 10 seconds.