The right answer isn’t about being right or wrong. It’s about what works for your users and what your team needs to achieve. Before you can choose “Buy” or “Purchase,” you need a clear strategy.
Strategy simply means getting everyone aligned on a shared focus. It tells the team what to do, and what to ignore.
A simple strategy statement connects four key elements:
- User Type
- User Need
- Value Provided
- Business Benefit
For example:
“We help budget-conscious buyers easily acquire high-quality audio gear so they can enjoy superior sound immediately, which supports AudioFlow’s goal of increasing first-time conversions and positive word-of-mouth.”
This statement immediately provides direction. Since the goal is easy acquisition for a wide, budget-conscious audience, choosing the clearest, most conversational word “Buy” over the more formal “Purchase” is the right strategic move.
A framework for writing with AI
The fundamentals of good writing are clarity, concision, and empathy, and they remain essential even with AI tools. While AI can draft text very quickly, its output often lacks soul. What makes UX writing truly effective is your ability to go beyond that basic text by understanding the user’s context, establishing a consistent brand voice, and skillfully modulating the tone.
To integrate AI responsibly without compromising your voice, ethics, or craft, adopt a mindset where AI is a co-pilot used for generating options, while you the designer remain the pilot responsible for strategy and empathy.
Use this three-step strategic filter to integrate AI outputs into your workflow:
1. Prompt with strategy
Your input to the AI must be dictated by your established voice and tone documentation, not just your immediate need. Instead of a simple request like “Write an error message,” let your prompt provide a clear strategy like:
“Write a toast notification message for a mobile app. The user just clicked an expired password reset link. Use a confident, empathetic, and direct tone. The message must tell them to request a new link, and it must be concise.”
This single prompt bundles the following elements:
- Voice & Tone: “confident, empathetic, and direct tone”
- Constraint: “toast notification message for a mobile app,” must “be concise,” and must tell them to “request a new link.”
The resulting AI output, “Your link has expired. Please request a new link to continue,” is a direct result of these layered instructions, showing how strategic input creates high-quality, on-brand UX copy right away.
2. Filter for empathy (Review)
The first and most critical step when reviewing AI output is to filter for the human element and inclusivity. AI outputs often rely on aggregated data that can perpetuate bias, use idioms, or assume an emotional state.
- Check for Exclusion: Does the output use any terminology (gendered, ableist, or overly colloquial) that might alienate a specific population?
- Check for Assumption: Does the message assume the user is happy, smart, or calm? (e.g., avoid “Great job!” or “Finally!”)
- Check for Idioms: Is the language literal and ready for translation, or does it rely on cultural shorthand that an international user or machine translator would miss?
3. Edit for core principles (Output)
After the output passes the empathy filter, focus on refining the language to ensure it is clear, concise, and authentically human.
- Clear: Edit for ambiguity. Ensure the language precisely describes the action and the object. For example, does the button label “Save” mean “Save All” or “Save this section”? This clarity is paramount.
- Concise: Cut out extra words. Look at what the AI wrote and remove anything you don’t need, like long, overly nice phrases or confusing descriptions. These just slow the user down. Your main goal is to get to the point quickly so the user can finish their task fast and easily.
- Human: Finally, edit for rhythm by reading the final output aloud to check for natural cadence. Does the writing sound like a robot, or does it flow naturally with appropriate contractions and pacing? The human ear is the ultimate tool for catching stilted or forced language, ensuring the final tone is authentic and engaging.
By following this three-step framework of strategic prompting, empathy filtering, and editing for clarity, concision, and human tone, you can use AI to create high-quality UX copy without losing your voice or affecting the user experience.
Designing error messages that help users
Error messages are often the most visible things we write, and they appear exactly when users are stressed and confused. While the best error is one that never happens, we must treat any error as a chance to help people finish their work.
When people use a product, they are focused on completing a task, not on figuring out how the app works. Running into an error is a hard stop to their goal. Instead of assuming the user is doing something “wrong,” your job is to design a better experience. As expert Kathy Sierra says, focus on helping your users become “more skillful, more powerful users [3].”
The problem with generic errors
Companies often blame the user for errors as a way to protect their own image. However, error messages that quietly point the finger at the user and make them feel bad actually harm the brand, causing people to stop using the product and decreasing loyalty.
When the real problem is a technology or design failure, you should be honest. For example, the email service Mailchimp uses its distinct brand voice to be apologetic, transparent, and helpful when an error occurs. This approach shows they’re committed to guiding the user forward, not blocking them.
If something goes wrong, do everything you can to make sure your users aren’t taking the blame. Focus on helping them achieve their goal.
The designer’s error framework
To guide users through or around failures, you must design your language using three principles:
- Avoid: The best error message is the one the user never sees. Use visual design and interaction (like autocomplete or automatic formatting) to prevent the error from happening in the first place.
- Explain: When an error occurs, clearly and quickly tell the user what went wrong and why (e.g., “Your return date must be after your departure date”). Never leave the user guessing.
- Resolve: Easily the most critical component. Always provide a clear next step. Don’t just say “File too large”; say, “This file exceeds the 10MB limit. Try compressing the file or uploading a smaller version.”
Your final task is to ensure the user is supported, not stalled, as they move toward their goal.