Verified against ChatGPT · 2026-08-14
Write microcopy for one flow's buttons, empty states, and confirmations as a single consistent voice
Writes all the microcopy for one flow at once — buttons, empty states, confirmations, and tooltips — as a single consistent voice pass, catching inconsistencies that appear when each string is written in isolation.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Write the microcopy for every touchpoint in the flow below at once, as one voice pass — buttons, empty states, confirmation messages, and tooltips. Microcopy is usually written string-by-string as each screen gets built, which is exactly how a product ends up with three different tones and two different ways of saying "delete" across one flow; writing it all together is how that gets caught.
FLOW AND EVERY STRING NEEDED
A shared-document flow: 'Share' button, empty state for a document with no collaborators yet, confirmation for removing a collaborator, tooltip on the 'view-only' permission toggle.
BRAND VOICE REFERENCE
Warm and plainspoken, avoids corporate jargon, occasional light humor in low-stakes moments, never uses exclamation points in more than one string per screen.
USER'S EMOTIONAL STATE AT KEY MOMENTS
Slight anxiety when removing a collaborator (worried about hurting the relationship or losing shared work); mild uncertainty at the empty state (not sure if sharing is even worth setting up yet).
RULES
Write every string in the same voice, established by the brand voice reference — but adapt tone within that voice to the emotional context at each moment; the voice stays consistent (word choice, formality, use of humor or restraint) while tone shifts appropriately (a delete-confirmation for a destructive action should not carry the same playful tone as an empty-state illustration caption, even in an otherwise playful brand voice). For every destructive or hard-to-undo action, the confirmation copy must state specifically what will happen and whether it can be undone — never a bare "Are you sure?" with no information added. For every empty state, the copy must tell the user what to do next to fill it, not just describe the absence ("No items yet" alone is a description; "Add your first item to get started" is actionable). Check every string against every other string in the set for consistency of terminology — if one button says "Remove" and another says "Delete" for functionally the same action, flag it and pick one term to use throughout. Keep every string as short as it can be while still being specific — cut filler words, but never cut the specific information (what will happen, whether it's reversible) needed for a confirmation to actually inform a decision.
WHAT NOT TO DO
Do not write generic placeholder-style copy ("Click here") that a template would produce regardless of context. Do not use exclamation points or forced enthusiasm on error or destructive-action copy even if the brand voice is generally upbeat elsewhere.
OUTPUT FORMAT
A table: Location in flow | String | Purpose (button / empty state / confirmation / tooltip). Follow with a short 'consistency check' note listing any terminology conflicts you found and resolved across the set.Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Microcopy generated string-by-string, in separate requests or separate design sessions, is the actual root cause of the terminology drift real products ship with — one screen calling an action "Remove" and another calling the same underlying action "Delete" isn't usually a deliberate choice, it's what happens when no single pass ever compared the two strings against each other. Generating the entire flow's copy in one batch, with an explicit cross-string consistency check as its own instruction, is what catches this, because the model is holding every string in context simultaneously and can be made to compare them directly rather than writing each one fresh without memory of the others. The voice-versus-tone distinction — consistent word choice and formality throughout, but tone that shifts to match the emotional weight of the specific moment — matters because a brand voice applied uniformly without this distinction produces a genuinely bad result in high-stakes moments: an upbeat, playful voice used verbatim on a destructive-action confirmation reads as tone-deaf regardless of how well it fits the brand everywhere else, and GPT-5.1 asked only for "copy in this brand voice" with no tone-modulation instruction will tend to apply the voice sample's surface style uniformly rather than recognizing that restraint is itself sometimes the on-brand choice for a specific moment. Requiring confirmation copy to state the specific consequence and reversibility rather than a bare "Are you sure?" addresses a well-documented UX writing failure where the most common confirmation dialog in software gives the user no actual information to decide with — "are you sure" answers nothing about what happens if they say yes, and a user who doesn't know whether an action is reversible has to guess, which is precisely the moment microcopy exists to prevent.
What you get back
Location: Remove-collaborator confirmation | String: 'Remove Jamie from this document? They'll lose access immediately, but any comments they've already left will stay.' | Purpose: confirmation. Consistency check: original draft used 'Remove' on this screen and 'Delete' on the collaborator-list row menu for the same action — standardized on 'Remove' throughout since 'Delete' implies destroying the person's account access history, which isn't what happens here.
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — branding & design, built and maintained for you — that's Scult's day job.
EXPLORE BRANDING & DESIGN
