E-commerce
Every hero, pricing page and funnel.Measured, not uploaded.
Real funnels, freshly measured
Browse by what you need:Funnels →Pages →Sections →Elements →
1790 funnels have a classified page structure. Newest shown.
What this is
Broken into parts you can actually use.
Every funnel we walk is split into whole pages, the sections inside them, and the individual elements (price anchors, ratings, guarantees) cropped to their exact measured pixels. Each one carries its receipts: the tactics detected, and the on-page string that fired. Point an agent at it and it pulls measured references before it designs; you review what it built in the Design Factory.
2421
Funnels walked
3638 pages screenshotted across them
1790
Landing pages measured
the entry page of each funnel, section by section
8396
Sections harvested
cut from those landing pages: heroes, pricing, FAQ, proof
1249
Elements cropped
5 types so far: price anchor, no credit card, rating, guarantee, countdown
How the scan reads
Three steps, one instrument
01
Point it at a URL
Any public lander, product page, or checkout. No login required.
02
It walks the funnel
Up to 4 steps (lander, product, cart, checkout) screenshotted and measured at each one.
03
You get the readout
Every tactic tied to its evidence string. No LLM opinion, no 0–100 vibes: deterministic, or absent.
a goldmine
– Pointed at our API, asked for nothing. This was its first message.Two independent sessions, same day. Not customer demand.
What each tool actually measures
No incumbent runs the full combo: each owns one slice. This is ours, honestly ruled off against theirs.
| UXXRAY | BuiltWith | Project Wallace | LLM page auditors | |
|---|---|---|---|---|
| Walks the full funnel, not one page | ✓ | ✗ | ✗ | ✗ |
| Every claim cites its evidence string | ✓ | ✗ | ✗ | ✗ |
| Deterministic, zero LLM opinion | ✓ | ✓ | ✓ | ✗ |
| Ships an agent-ready build brief | ✓ | ✗ | ✗ | ✗ |
Candidly
Frequently asked, honestly answered
What do I actually do with this?
Point your AI at a measured funnel before it designs anything, not after. Hand it a report URL or the library API and it works from real tactics and proof language pulled off a live page, not a guess.
Is it free?
Yes. Every scan grows the 2,421-funnel corpus you can compare against, so giving it away early makes the reference set more useful for the next person who runs one.
Do I need an account?
No. You can scan a funnel or browse the library without signing up for anything. If you do give us an email, to unlock a playbook or get a finished teardown sent over, we use it for that and occasional product updates. Never sold, never shared.
How do I use the library in my own projects?
Fetch /llms.txt for the machine-readable map of every endpoint, or call the library API on api.uxxray.com directly for specimens with measured evidence attached. Prefer to browse instead of code? The same library is right there, visually.
Are the teardowns real?
Yes. Every fact is measured live from a page's own DOM and computed CSS at crawl time: no LLM guessing, nothing hand-typed. Open any report in the hub and check any claim against the live site yourself.
Is this legal?
Yes. We only read what's already public: rendered HTML, computed CSS, and visible copy on pages anyone could open in a browser. No logins, no paywall bypassing, no scraping behind auth. Automated crawls identify as UXXRAYBot (see /bot.html) and honor robots.txt and other crawler directives.
What happens if a site blocks bots?
We label the result "blocked," never zero. Cloudflare challenges and login walls happen. We say so instead of guessing a score.
X-ray a funnel