Microcopy at the Point of Action

The last thing a visitor reads before clicking a button is usually about five words long, and those five words carry more weight per letter than anything else on the page.

I mean the button label itself, and the little line under it, and the note beside the credit card field. Designers call this microcopy, which makes it sound decorative. It isn’t. It’s the text stationed at the exact spot where a decision gets made, and decisions have a peculiar property: they concentrate doubt. A visitor can read your whole page in a state of relaxed agreement and then, hovering over the button, suddenly remember every question they have. What happens after I click? Will this cost money? Am I signing up for spam? Can I undo this? The page had thousands of words to answer these questions, but the visitor isn’t going to scroll back up. Whatever is within an inch of the button is the only text still in play.

This is why generic button labels are such a waste. “Submit” answers none of the questions doubt is asking. It describes the mechanics of the form, from the form’s point of view. Nobody wants to submit. People want to get the report, start the trial, book the call, see the price. A label written from the visitor’s side of the glass — naming what they receive, not what the database does — converts the button from a demand into an offer. The test I use is simple: does the label complete the sentence “I want to…”? “I want to submit” fails. “I want to see my results” passes.

The line under the button deserves equal care, because it’s where you neutralize the specific fear that the label raises. “Start free trial” raises the fear of a forgotten subscription silently converting into charges, so the line under it should say whether a card is required and what happens when the trial ends. “Book a demo” raises the fear of being handed to a salesperson forever, so the line should say how long the call is and who’s on it. Notice the pattern: the reassurance has to match the anxiety of that particular action. A generic trust badge doesn’t do this. The words “no credit card required,” in small quiet type, under exactly the button where the visitor is wondering about their credit card, do more than any badge because they answer the question actually being asked at that moment.

There’s a discipline to this that resembles good engineering: put the information where it’s needed, not where it’s convenient to store. Your refund policy lives on a policy page, and that’s fine as the canonical source. But the moment someone is deciding whether to pay you is the moment the refund policy is load-bearing, so a one-line summary of it belongs next to the pay button, the way an error message belongs next to the field that caused it. When I built the checkout flow for GazeSite, the changes that mattered were embarrassingly small: saying near the button what would happen immediately after payment, and what to do if it didn’t. Tiny sentences. But the point of action is where tiny sentences do their work.

Microcopy also has a failure mode worth naming: cleverness. There’s a fashion for jokey button text and for the manipulative variant of it, the decline link that says something like “No thanks, I prefer failing.” The first is a gamble, because a joke asks the visitor to spend attention decoding it at the precise moment their attention is scarcest. The second is worse than a gamble; it’s an insult with a click target, and visitors correctly read it as a company willing to mock them to make a number go up. Whatever it gains in one metric it pays back in the visitor’s overall estimate of your character, and character is what they’re really evaluating when they decide whether to hand over an email address or a card number. Plain and honest beats clever almost everywhere on a page, and at the point of action it beats clever always.

One more place microcopy earns its keep: the moments after something goes wrong. A payment fails, a form rejects an input, a page can’t be found. These are points of action too, just involuntary ones, and the default text there tends to be written by whichever system threw the error. “An error occurred” is the microcopy equivalent of a shrug. The visitor at that moment needs three things in one short line: what happened, whether it was their fault, and what to do next. Companies polish the happy path and let the sad paths speak in stack-trace dialect, which is exactly backward, because the sad paths are where a visitor decides whether you’re worth a second attempt.

If you want to audit your own site for this, ignore the paragraphs and read only the buttons, the labels, the placeholder text, and whatever sits within a thumb’s reach of each commitment. Read them as a stranger who is mildly suspicious and slightly in a hurry, because that’s who your visitor is. Every place the small text is vague, doubt fills the vacuum. The paragraphs above the fold are your speech. The microcopy is your handshake. People forgive a mediocre speech. They remember the handshake.