A product team is redesigning their checkout flow and asks AI to write the copy for each screen. They provide the screen names, the actions users need to take, and the information fields. What comes back is clean and accurate: button labels that name the action, field labels that name what to enter, error messages that name the error. Technically correct. When they put it in front of users in testing, two things keep happening: people hesitate before entering their payment information, and three out of eight testers abandon at the order review screen because they are not sure they can change something before it is finalized. The copy described what each screen does. It did not address what users are worried about when they get there.
Good UX writing anticipates the question before the user asks it.
Why Descriptive Microcopy Misses Friction
UX writing has two jobs: it names what is happening (the label, the action, the state), and it reduces friction at the moments where users hesitate or lose trust. The first job is about accuracy. The second is about psychology — specifically, the user’s mental model at each point in the flow and what they need to know to proceed with confidence.
Most AI-generated UX copy does the first job well and the second job not at all, because the second job requires knowing where users hesitate, what they are worried about, and what specific piece of information would dissolve that worry and move them forward. That knowledge comes from user research — from watching people use the product and noting the exact moments where they pause, re-read, or abandon. Without that research in the brief, AI produces descriptively accurate copy that does not address the real friction.
What a UX Writing Brief Needs to Capture
A UX writing brief that addresses friction needs three inputs that most copy briefs omit.
The user’s mental model at each screen: what does the user believe is happening, what are they trying to accomplish, and what are they uncertain or worried about at this specific point? The mental model at the checkout payment screen is different from the mental model on the order review screen — and the copy needs to address the specific concern at each point.
The known friction points: what do user testing or analytics show about where users hesitate, re-read, or abandon? These moments are the locations where the copy has the most work to do. The copy brief should specify each friction point and the specific worry that the copy needs to resolve.
The information that reduces the specific worry: what does the user need to know to proceed confidently? This is often not the same as what the screen is describing. An order review screen should include something that confirms “you can still change this” — not just a description of what the order contains.
What a Properly Briefed UX Copy Request Looks Like
Role: You are a UX writer helping write microcopy for a redesigned
e-commerce checkout flow.
User context: Customers checking out for the first time on this site.
They are familiar with standard checkout flows but have been burned
before by checkout processes that charged them before they were ready
or made it difficult to correct mistakes.
Friction points and the copy's job at each:
1. Payment information screen:
User worry: "Is this secure? Will my card be stored without asking?"
Copy job: Reassure on security and be explicit that storage is
optional and opt-in. The security badge is present but copy should
reinforce it, not just repeat it.
2. Order review screen:
User worry: "If I submit this, is it final? Can I still change
the address or cancel?"
Copy job: The review screen headline and subtext should make it
explicit that this is still the review stage, that changes are
still possible, and what happens when they place the order.
Users are hesitating because they don't know if "Place Order"
is reversible. The copy should answer that before they ask.
3. Confirmation screen:
User worry: "What happens now? Did it actually go through?"
Copy job: Immediate confirmation of what was ordered, what
happens next (shipping confirmation email timing), and what
to do if something looks wrong. Users need to know what
"done" actually looks like.
Write copy for each screen that addresses the specific worry,
not just the action. Button labels and headers are the priority.
Supporting copy where needed to resolve the friction.
The copy from this brief works at the points where users actually hesitate. The order review screen tells users they can still change things before they have to wonder. The payment screen addresses security before the user’s uncertainty triggers an exit.
The User’s Worry Is the Writing Prompt
Every piece of UX copy that reduces friction was written in response to a specific user worry at a specific moment in a specific flow. The brief that contains those worries — sourced from user research, testing, or analytics — gives AI the actual prompt that each piece of copy needs to respond to. UX writing without that brief produces labels and descriptions. UX writing with it produces guidance that moves users forward because it answered the question before they stopped to ask it.
For product teams and UX writers building or redesigning interface copy, Briefing Fox structures the brief so user mental models and friction points are captured before any microcopy is written.
Before Your Next Interface Copy Project
Before asking AI to write any UX copy, identify the one screen or moment where users most often hesitate or abandon — and write down what they are worried about at that moment. That worry is the brief for that screen’s copy. The microcopy that reduces friction is written in response to the user’s question, not in description of the screen’s function.
Try Briefing Fox free at www.briefingfox.com.
Descriptive copy names what a screen does. Functional copy reduces the friction that causes users to hesitate, re-read, or abandon at that specific screen. Good UX writing does both — but the functional work is what reduces drop-off, and it requires knowing where users actually hesitate.
Look at your analytics for the screens with the highest abandonment or the lowest completion rates. Then watch a usability recording of a user at that screen to see what they do before they leave. The hesitation you observe is the friction the copy needs to address.
The user’s mental model at the specific screen (what they believe is happening and what they’re worried about), the known friction points from testing or analytics, the information that would resolve each hesitation, and the action the user needs to take with confidence. These inputs shift the copy from labeling to guiding.
Brief AI with the specific error, what caused it, what the user can do to fix it, and the user’s emotional state when they encounter it. An error message that names the problem, explains the cause in plain language, and offers a specific next action is fundamentally different from one that names the error code and leaves the user stuck.