The checklist is in English so it's easy to copy directly into your AI tool of choice. Adapt it to your product and language as needed.
Layer 1 - Foundations: rules, constraints, and deviations
Build this once per product. Update only when something structural changes (a new regulation, a new platform, a company-wide style shift).
Regulations & compliance
•	[ ] List every regulation your product has to comply with (finance, health, accessibility, children's services, gambling, anything)
•	[ ] List phrases you're required to use without changing anything - disclaimers, consent language, mandated legal terms
•	[ ] List phrases you're forbidden from using - "guaranteed," "risk-free," specific medical claims, anything legal told you to drop
Brand language rules (deviations from standard UX writing)
•	[ ] Sentence case or title case for buttons and headings?
•	[ ] Oxford comma - always or never?
•	[ ] Exclamation marks - never, few, or many?
•	[ ] Link to your company-wide style guide if you have one (pull out the rules that matter for writing, not the brand-deck rules about logo padding and typography)
•	[ ] Terms that have fixed formatting - capitalization, hyphenation, one word vs. two ("sign in" vs. "sign-in")
Plain language & reading level
•	[ ] Target reading level (rough range is fine: "8th grade consumer," "technical professional")
•	[ ] Technical terms users already know and you should not soften (API, deductible, escrow - whatever your audience knows)
•	[ ] Internal terms that should never leak into the UI (codenames, backend field names, engineering shorthand)
Technical constraints
•	[ ] Character limits per component - buttons, headings, tooltips, mobile vs. desktop
•	[ ] Dynamic content ranges - currency amounts, date formats, number formats, units
•	[ ] RTL or multi-language considerations if relevant (expansion for some languages, pluralization, gender)
Layer 2 - Brand & voice
Build this once per product. Update when the brand repositions or the voice strategy shifts.
Brand values
•	[ ] Three to five values in your own words, with a short explanation of each value
•	[ ] One brand you admire for their writing - and specifically what they do that you want to learn from. You can also include some sentences as examples
•	[ ] One brand whose writing you'd avoid sounding like - describe and specifically what they do wrong
Voice & tone
•	[ ] Three voice adjectives - each with its opposite and an explanation of what it means exactly ("direct, not blunt, because users need clarity but shouldn't feel talked down to")
•	[ ] Humor - yes or no? If yes, what kind, and in which contexts?
•	[ ] Formality rules: contractions yes/no, slang yes/no, first name when known yes/no, user address ("you" / "me" / both), pronouns
•	[ ] Tone modulation by context - how the voice shifts for errors, milestones, loading states, time-sensitive actions
Terminology
•	[ ] What you call your users ("users," "members," "pros," something brand-specific) - and why
•	[ ] Terms you chose deliberately over a common alternative - and the reason
•	[ ] Competitor terms you deliberately avoid - and the reason
Product description
•	[ ] What the product is - in one paragraph
•	[ ] What the user gets from it - in one paragraph (value, not features)
•	[ ] Two or three things you do that competitors don't, or don't do as well
•	[ ] The "aha moment" - when does the value click for the user?
User pain points
•	[ ] What pain this product solves - specific person, specific moment, specific feeling
•	[ ] What users try before finding you - what they're working around or giving up on
•	[ ] What they feel after using your product for the first or tenth time
Layer 3 - Per-flow context
Create a version of this for every flow or screen you're writing. This layer captures what's specific to this moment in the product.
Flow identification
•	[ ] Flow name (specific - "first screen after signup," not "onboarding")
•	[ ] Scope: major flow or small piece (error state, tooltip, empty state, confirmation)
Desired emotion
•	[ ] What you want the user to feel during the flow (specific - "relieved that their data is safe," not "good")
•	[ ] What you want them to feel after the flow ends
•	[ ] What emotion you're specifically trying to avoid (confused, anxious, rushed, patronized, overwhelmed)
User life situation
•	[ ] What's happening in the user's life at this exact moment (situation, not demographics)
•	[ ] What specific thing they're afraid will go wrong (specific fear, not "making a mistake")
•	[ ] Physical and attention context - where are they, what device, what else is competing for their attention
User expectations
•	[ ] What they expect to happen in this flow before they start it
•	[ ] Expectations they have that are wrong - things your product doesn't do, or does differently
•	[ ] Surprises in this flow, good or bad - things they won't see coming
•	[ ] What would feel like a broken promise - things the UI should never promise
User journey
•	[ ] What they were doing just before this flow
•	[ ] What brought them to the product today (push notification, email link, opened the app, deep link)
•	[ ] Ideal next step after this flow ends
•	[ ] Failure mode - where they go if this flow fails (bounce, support ticket, competitor, churn)
Tip to make it easier: You can give the 3rd layer to an AI, then record an answer to all questions and have it organize the information for you.

