Every few weeks someone sends us a link and the same sentence: "the site looks great, but nobody signs up." Last quarter we ran eleven of these audits. Two of the pages genuinely needed a redesign. The other nine were fine — clean type, decent hierarchy, fast enough.
The pattern we kept finding had nothing to do with how the page looked.
The page answered the wrong question
Almost every underperforming page we opened was built to answer "what is this product?" The visitor arrived asking something else entirely: "is this for someone like me, and what happens if I click?"
Those are different pages. The first one explains features. The second one removes doubt. When you write the first and expect the second to happen, conversion sits at 1% and no amount of gradient fixes it.
Nobody bounces because your spacing was off. They bounce because they could not tell whether it was for them.
Three things we changed first
Before touching layout, we rewrote the same three places on every page. In most cases that alone moved the number enough to justify the audit.
1. The subheadline names the reader
Headlines get all the attention and most of the budget. The line under it does the real work: it says who this is for and what changes for them. "Analytics for teams" becomes "See which of your ads actually pay, without a data analyst."
2. The button says what happens next
"Get started" is not a promise, it is a shrug. "Start free — no card" is a promise. People hesitate over buttons far longer than designers assume, and the label is the only information they have at that moment.
3. The first objection appears above the fold
Every product has one question that stops people cold: price, setup time, lock-in, migration. Burying it three sections down does not delay the doubt, it just sends the reader to a competitor to find the answer.
What we stopped recommending
- Longer pages. Adding sections to a page nobody finishes is a way to feel productive, not to convert.
- Social proof without specifics. "Loved by thousands" reads as decoration. One customer sentence with a number reads as evidence.
- Animation on the primary path. A 400 ms reveal between the reader and the signup button costs more than it adds.
So when is it actually design?
Two of the eleven pages did need rebuilding — both were product dashboards pretending to be marketing sites, with navigation nobody could parse. That is a real design problem and no copy edit saves it.
But if your page is readable, loads fast and still converts under 1%, start with the words. They are cheaper to change and they tell you what to redesign.