Voice and tone
Five questions every design system answers about voice and tone. These are Dense data’s answers, and each one points back to the reasons that chose it.
View all 5 voice and tone tokens ↓The reasons this scheme answers from
- R1Business and productRuns daily operations, one
- R2People and work contextDaily experts, AA
- R3Team and deliveryA small team, every month
- R4Tasks and flowsScan and compare, constantly
- R5BrandPrecise, quiet
- R6OutcomesTime on task, acting on the wrong record
Question 1
Which kinds of copy does it need?
- Labels and messages (chosen)
- Labels, messages and help
- Labels, messages and articles
- Labels, messages and headlines
Why Dense data picks it: Labels and messages are all the copy an operations tool needs.
becauseR1
Question 2
How many tones does it use?
- One (chosen)
- One per moment
- One per brand
Why Dense data picks it: One flat tone everywhere; experts want no surprises.
becauseR2
Question 3
How is the voice defined?
- Rules and a word list (chosen)
- Principles and examples
- A shared core, brand tone on top
Why Dense data picks it: A word list keeps labels identical across screens.
becauseR3R4
- Rules and a word list
These tokens are the decisions themselves; no primitive sits under them.
Question 4
How does it sound?
- Terse (chosen)
- Plain and direct
- Conversational
- Playful
Why Dense data picks it: Experts want short labels and short messages.
becauseR2R5
- action
- commit
- Save
- cancel
- Cancel
- destroy
- Delete
- retry
- Retry
- empty-action
- Change filters
Question 5
Where do the words matter most?
- Errors and empty states (chosen)
- Articles and guides
- Form steps and errors
- Headlines and buttons
Why Dense data picks it: Errors and empty tables are where words decide the next action.
becauseR6
- button
- input
- overlay
- table
When to use
- A button labelUse: One verb from the word list. Not: A sentence or "Please". Labels repeat on every screen; one verb reads fastest.“Export”
- The words of an errorUse: Name the record, then the fix. Flat and specific.. Not: "Oops" or a vague apology. Experts act on the message; the record and the fix are all they need.“Alert ALR-2042 not saved. Threshold must be between 0 and 100.”
- A success messageUse: A short confirmation with the count. Not: Celebration. Confirm and get out of the way; the next task is waiting.“12 alerts exported.”
- An empty stateUse: Why it is empty and the one action that fills it. Not: A blank table. An empty table with no words looks like an error.“No escalated alerts. Change the date range to see older ones.”
- A confirmation questionUse: Name the exact object and the consequence. Not: "Are you sure?". Acting on the wrong record is the costly mistake.“Dismiss 3 alerts? This cannot be undone.”
- A first visitUse: No tour. Show the data, with a tooltip where needed.. Not: A welcome tour. Daily experts skip tours; tooltips cover the first uses.“Hover any column header to see what it counts.”
All voice and tone tokens
5 tokens · each decision and the primitive it points to
- Decisions · 5
content.action.commitSavecontent.action.cancelCancelcontent.action.destroyDeletecontent.action.retryRetrycontent.action.empty-actionChange filters
Decisions reference primitives, so changing one primitive updates every decision and component that uses it.
Voice and tone in Dense data: terse. Experts want short labels and short messages. Errors and empty tables are where words decide the next action.