Teardown: How I Score a Vibe-Coded Landing Page in 60 Seconds
The exact read I run before I have consciously thought about it. You can steal it.
July 20, 2026
When a landing page loads, I have a number before I have a thought. Here is the sixty second read that produces it, broken into the steps I do not usually notice myself doing. Run it on your own page and you will see what a stranger sees.
Second 1: the pattern flash
Before I read a word, my eye matches the page against every page it has seen. Gradient hero, glass cards, three feature columns, and the flash returns "seen it." That is the fifty millisecond trust read, and it is most of the score right there.
Seconds 2 to 10: can I tell what this is
I look for one sentence that names the exact thing this does for an exact person. If the hero says "build the future," the score drops, because a vague headline is a tell that the founder outsourced the one decision that mattered. If it names a real job, points go back up.
Seconds 10 to 30: the tell sweep
Now I count. Purple gradient. Inter on everything. 16px radius uncustomized. Emoji standing in for icons. No empty state visible in the screenshots. This is the twelve tell checklist, run fast. Each hit is a point of slop.
Seconds 30 to 60: is there a human here
Last, I look for proof a person stands behind it. A face, a real name, a specific number, a screenshot of the product actually working. Its absence is the quiet killer. Anonymous plus generic reads as risk, and risk loses the signup.
The score
Add the tells, weigh the trust read, and you land in a band. Clean, mild, or heavy, the same three bands the 1,590 launch study used. Most vibe coded pages land in mild, six or so tells, and every one of them is a ten minute fix.
The reason you cannot run this on yourself is simple: your pattern flash is gone. You have seen your page a thousand times. So borrow a fresh sixty seconds. Run your page through auditmy.design and get the read as a number and a pinned list.
Want yours torn down in a post? Reply with the URL.