The Craft of Micro-Copy

Typography

The difference between Delete and Remove, why I rewrite placeholder text ten times, and how a single word change lifted conversion by 12%.

There is a moment in every project where the big decisions have been made — the layout is locked, the color palette approved, the interaction patterns finalized — and you are left staring at a button that says "Submit." That single word, sitting inside a 120-pixel-wide rectangle, will be the last thing a user reads before they commit to an action. And yet, in most design processes, it receives less attention than the border-radius of the card it sits on. I have spent the last eight years obsessing over these tiny fragments of language, and I have come to believe that micro-copy is not a detail. It is the design.

The difference between "Delete" and "Remove" is not semantic trivia — it is an emotional contract. "Delete" implies permanence, destruction, finality. "Remove" suggests separation, reversibility, gentleness. When I was redesigning the file management system at my previous company, we ran a five-second test with both words on the same destructive action. Users who saw "Delete" hesitated an average of 3.2 seconds longer and were 23% more likely to cancel the action entirely. Users who saw "Remove" completed the task faster but later reported confusion about whether the file was truly gone. The right word depends on the truth of what happens next, and the designer's job is to find the word that is both honest and humane.

Image

Copy iterations for a single modal — attempt 14

I rewrite placeholder text at least ten times before shipping. This is not perfectionism — it is process. The first draft of placeholder text is always too clever, the second too vague, the third too long. By the seventh or eighth attempt, I start to find the rhythm of the interface: the cadence that matches how a user actually thinks when they encounter the field. For a search bar in a recipe app, I went through "Search recipes," "Find something to cook," "What are you craving?", "Search by ingredient or dish," and six other variations before landing on "Tonight's dinner starts here." That version tested highest for engagement because it spoke to intent, not mechanics. I keep every iteration in a spreadsheet — one row per string, columns for context, character count, test results, and the date I last revised it. That spreadsheet is the most important design artifact I own.

Empty states are the most underrated design surface in any product. When a user encounters an empty list, an empty inbox, or a blank canvas, they are at their most vulnerable. They do not yet understand the product's value, or they have just completed a task and need reassurance. The copy in that moment sets the emotional tone for everything that follows. I once changed an empty state message from "No results found" to "Nothing here yet — try adjusting your filters or broadening your search" and watched the retry rate climb from 12% to 31%. That single paragraph of copy did more for retention than the three features we shipped that quarter. Empty states deserve the same rigor as onboarding flows, because in many cases, they are the onboarding flow.

Image

The spreadsheet where all my micro-copy lives

The most dramatic result I ever achieved with a single word change happened on a pricing page. The call-to-action button read "Buy Now." We changed it to "Start Now." Conversion increased by 12% in the first week and held steady over the following month. The insight was simple in retrospect: "Buy" centers the transaction, "Start" centers the experience. Users were not buying software — they were starting a new chapter of their workflow. That shift in framing, accomplished with a five-letter word, was worth more than the entire redesign of the pricing tier cards. This is why I believe designers should write their own copy rather than handing it off to a separate copywriting team. The person shaping the interaction must also shape the words inside it. Language and layout are not parallel tracks — they are the same track.

Micro-copy is where trust lives. Users may not consciously read your button labels and error messages, but they feel them. Every word that matches their mental model builds invisible trust. Every word that doesn't creates invisible friction. The smallest strings carry the largest emotional weight.

"The best interface copy is invisible. You only notice it when it's wrong."

My process has become ritualistic: I write all micro-copy in a dedicated spreadsheet before it ever enters a design file. Each string gets tested with five-second tests — I show a user the screen for five seconds, hide it, and ask them what they think the button or message said. If they cannot recall it accurately, the copy is too clever. If they recall it but misunderstand the action, the copy is misleading. Only when recall and comprehension align do I consider the string ready. Designers often resist this level of rigor for "just words," but those words are the voice of your product. They deserve the same iteration cycles as any visual element.

Does this word match the user's mental model of the action?
Can the user predict the outcome from reading this label alone?
Is the tone consistent with the rest of the interface?
Is this the shortest version that still communicates clearly?
Would a non-native English speaker understand this without context?
Does the error message tell the user what to do next?
Have I tested this with a five-second recall test?
Does the empty state invite action rather than state absence?