Choose a link or button by what it does
By Gyata Tabiri · Revised November 2, 2026
A control makes a promise before anyone uses it. “View case study” promises a destination. “Send message” promises an action. I want the wording, HTML, and result to agree, even when both controls share the same visual style.
Appearance does not choose the element
A link can look like a button and still be a link. That is appropriate when it navigates to another page. Its destination can be copied or opened in another tab. Replacing it with a scripted button just to get the appearance can discard useful behavior without improving the task.
An action needs its own feedback
For a submission, I want to know what happens after activation: is the request still processing, did it fail, or did the server accept it? Changing the label to “Sent” before that response would give the interface a confidence the underlying process has not earned.
Small defaults can have large consequences
Inside a form, a button’s type needs attention. A secondary control should not accidentally submit the form. When reviewing the flow, I would operate it from the keyboard as well as click it, checking the result rather than assuming the intended behavior follows from the visual hierarchy.
Read the labels as a sequence
A page full of “Learn more” links may force someone to reopen the surrounding text to understand each destination. A useful label gives enough context to choose. It does not need to describe the entire page, but it should distinguish this next step from the others.
The visual treatment can make an action easy to find. The element and the behavior are what make it dependable once someone finds it.
Read alongside: Keeping a control’s state truthful · What happens after Send.
Reference: HTML button element.