UX writing is the practice of writing the words inside a product or page: buttons, form labels, hints, error messages, and the short reassurance lines next to them.
Those small pieces of text are called microcopy, and their job is to answer the question a visitor has at the exact moment they hesitate.
In our database, microcopy changes such as adding "No credit card required" or renaming a link from "Request a quote" to "Request pricing" have lifted sign-ups and clicks by double and even triple digits, while other wording changes made no difference or lost.
This page defines the terms, then walks through microcopy examples from real A/B tests, grouped by where the words sit on the page.
For the wider copy picture, see our copywriting hub.
What microcopy is, and how it relates to UX writing
The term comes from a June 2009 post by Joshua Porter, "Writing Microcopy".
Porter described it as "small yet powerful copy": "a short sentence, a phrase, a few words. A single word."
The example that started it was a conversion problem.
Checkout transactions on a project Porter worked on kept failing because people entered an address that didn't match their card.
Adding one line, "Be sure to enter the billing address associated with your credit card," made the errors go away, according to Porter's post.
UX writing is the wider discipline. A UX writer decides what every piece of interface text says, how it sounds, and when it appears.
Microcopy is the output of that work. Marketing copy, by contrast, is the headline and body text that sells; microcopy is the text that helps someone finish the action the marketing copy asked for.
Where microcopy lives
| Location | What the words do | Typical question it answers |
|---|---|---|
| Button and link labels | Name the action and what happens next | "What do I get if I click this?" |
| Text next to a button | Remove a last-second objection | "Will this cost me anything?" |
| Form labels and hints | Explain what to enter and what is optional | "Do I have to fill this in?" |
| Error messages | Say what went wrong and how to fix it | "Why won't this submit?" |
| Placeholders, nav labels, badges | Help people find things | "Where is the thing I want?" |
For error messages, Nielsen Norman Group's error-message guidelines (Tim Neusesser and Evan Sunwall, 2023) recommend showing the message close to the error's source, using human-readable language, offering constructive advice, not blaming the user, and preserving what the user already typed.
Mailchimp's content style guide adds two rules that matter for conversion: links should describe their action or destination rather than "click here" or "learn more," and required fields should be labeled clearly.
Why tiny text can move big numbers
Microcopy sits at the point of decision. A visitor who has read the headline, liked the offer, and reached the button is close to converting, and a single unanswered doubt is often what stops them.
That is also why microcopy tests usually measure a micro-conversion such as a click or a form step.
Keep that in mind when reading the numbers below: a big lift in clicks is not the same as a big lift in revenue.
Microcopy examples from real A/B tests
Each example below comes from our test database, with the numbers exactly as the team that ran it reported them.
They are grouped by where the copy sits.
1. Reassurance next to the button
The most consistent pattern is a short line that removes a fear about cost or commitment.
- "No credit card required." Teamleader added this line to its free-trial sign-up page. Sign-up conversion went from 48.7% to 54.8% in the Netherlands and from 33% to 38.6% in Belgium, reported as a 12.5% lift (Teamleader test).
- "It's free!" Soocial placed these two words beside its "Sign up" button. Click-through to the sign-up form rose from 14.5% to 18.6%, a 28% lift, though that measures clicks, not completed sign-ups (Soocial test).
- Data-privacy promise on a form. Groupama showed a message saying personal data was protected to one segment of visitors on an insurance quote form, and quote submissions rose 10% (Groupama test). It was targeted at a single segment, so don't assume the same line works for every visitor.
Reassurance works best when it answers a doubt your visitors actually have.
Oliver Kenyon wrote three reassurance lines from customer survey answers and placed them above a collection page grid, a test credited with $44,890/month (Oliver Kenyon test).
The lines came from research, not a template.
More on this kind of text in our guide to trust signals.
2. Button and link labels
Label tests have the widest swings, both up and down.
The common thread among winners is that the label tells people what they get, in their own words.
- "Request pricing" vs "Request a quote." Veeam changed only the link text on a product page. Click-through to the sales inquiry page went from 0.54% to 1.40%, a 161.66% lift. Veeam measured clicks to the inquiry page, not submitted leads (Veeam test).
- "Trial for free" vs "Sign up for free." Going reported a 104% rise in premium trial starts month over month. The button also sent people to a different offer, a premium trial instead of the free plan, so the words were not the only change (Going test).
- "Start registration" vs "Sign up today." Booking.com compared several labels on its partner registration button. "Start registration" did best and "Sign up today" cut clicks by about 10% (Booking.com test).
Clever wording often loses to plain wording. Dave Diederen tested "Add to Cart" against "Try Risk Free Today," "Get Yours Now," and "Start Your Journey," and "Add to Cart" won every time (Dave Diederen test).
Encyclopaedia Britannica replaced "Start your free trial" with "Become a subscriber" and "Become a member," and both variants lost, with click-through down as much as 20% (Britannica test).
Adding the price to the button text also lost for Dave Diederen, at -2.03% revenue per visitor (price on button test).
Our CTA copy hub collects every button-label test in the database.
@DaveDiederen3. Form labels, hints and optional fields
Form microcopy tells people what to enter and what they can skip.
Small changes here show up in completion rates.
- Required-field markers. Micro Focus added asterisks to mandatory fields on its free-trial form, plus a "required field" note above it. Form completions rose 5.89% over a 20-day, 50/50 test (Micro Focus test).
- "Optional" labels instead of asterisks. Springest went the other way: it labeled the optional fields instead, alongside other changes such as adding user quotes and brand logos. Completions rose 5.21% at 97% significance over 2,500+ conversions, but the changes were bundled (Springest test).
- Clearer labels. Qualicorp replaced hard-to-read field labels with more legible ones that appear above a field when it is clicked. Completed quote requests rose 16.93% in a test VWO reported (Qualicorp test).
- An escape hatch. Tinkoff Bank added a "Don't remember passport details?" checkbox that let applicants send them later. Applications rose 35.8% across 265,000+ visitors in 23 days (Tinkoff test).
The Micro Focus and Springest results point in opposite directions, and that is useful.
What matters is that people can tell which fields are required, not which marker you use.
See our form fields hub and form optimization guide for more.
4. Placeholders, navigation labels and helper text
- Labels on icons. Goat Milk Stuff added short text labels beside its mobile header icons for Menu, Search, Account and Cart. Conversion rate rose 10.37% and revenue per visitor 2.95% in a two-week mobile test run by Blend Commerce (Goat Milk Stuff test).
- Search placeholder. Blend Commerce rewrote the mobile search placeholder on Azuna to invite shoppers to search for products, and revenue per visitor rose 3%. The new wording wasn't published and no significance was stated (Azuna test).
5. When more words hurt
Microcopy is not free. Every line competes for attention, and a message that raises a new question can cost you.
Clean Commit replaced two shipping messages on product pages with one price-aware line, a free-shipping threshold reminder on cheaper items and "ships free" on pricier ones.
Conversion rate fell 5.5% and revenue per visitor 9.0% across 20,974 visitors, with a 75.1% probability that the control was better (Clean Commit test).
A threshold reminder can make a shopper notice they are short of free shipping.
How to test your own microcopy
- Start from a real doubt. Read support tickets, chat logs and survey answers to find the question people ask right before they quit. Porter's billing-address line came from error logs.
- Change one thing. If the label and the destination both change, as in the Going test, you won't know which one did the work.
- Measure past the click. Several winners above measured clicks only. Track the sign-up, lead or order too, since a label that promises more can lift clicks and then disappoint.
- Plan for small effects. Most wording changes move results by a few percent, which needs a lot of traffic to detect. Check with our sample size calculator before you launch, and read results with statistical significance in mind.
For page-level copy beyond the interface text, see our guide to landing page copy.