CTA button best practices: match the headline

You wrote the label after the copy, so it matches the product you know. A visitor reads the page in the order it appears: the headline, the sentence under it, then the button. If the headline asks whether they are losing cake orders to messages, and the button says Submit, the visitor has to guess what Submit does and whether it costs anything. On a phone the button may also sit below the first screen, or under a header that hides it while they scroll. A visitor who never clicks is often not refusing the offer. They never saw a label that finished the sentence they had just read.

• The label names the result the visitor gets, such as a free order page, not a generic verb.
• The button appears on the first phone screen, or a fixed one stays in view while they scroll.
• The button is filled, with contrast and padding that a text link does not have.
• One main action per screen, and the secondary link looks clearly secondary.
• A click changes the screen or shows a loading state within a moment.

The label works when a reader can say what it does without reading anything else. Take a headline that promises to stop lost cake orders. A label that says Submit gives no reason to press it. A label that says Start a free order page carries the promise forward and tells the visitor what happens next. Test the pair by reading the headline and the button one after the other, without the rest of the page. If you have to explain the button to yourself, a visitor will not manage it either. Keep the same nouns in both places, so the visitor does not have to translate between them.

Landing page testing

On a laptop, the button often sits in the first view with the headline. On a phone the first view is shorter, and a photo or a video can push the button below the fold. A sticky header can hide it while the visitor scrolls back up. Check the first screen on a phone, with the browser's bars showing, and see whether the button is there without scrolling. If it is not, add a second button after the first section, or make the header button the main action. The report shows the step where each visitor stopped, so you can see whether they stopped before the button or after it.

A visitor who presses a button and sees no change will press it again, then assume it failed. Common causes are a button that only checks the form after submit, a request that takes long enough for the page to look frozen, or a button that opens a form further down the page where nobody looks. Give the click an immediate result: a loading label such as Creating your page, a jump to the form, or a change of screen. Disable the button while a request runs, so one press makes one request. The test shows whether the click changed anything, and it records a slow or failed request on the step where it happened.

A headline test, a new price or a new ad angle changes what the button has to finish. The label can drift from the headline while both look fine alone. Rerun the same visitors after each change, on a phone and a laptop, and check they read the button as the next step. Each run can also set how visitors arrive, a search keyword or an ad, so you can check the label against that source.