Experience System

Voice & Tone

How our products communicate—clear, professional, confident, human, and concise. One personality across every product and tool.

The golden rule: Clarity always wins over personality.

Voice vs. Tone

Voice is who we are—clear, professional, confident, human, concise. Our voice is consistent across every message.

Tone is how we adjust for the context—the same voice. It is warmer in a welcome, calmer in an error, and shorter in routine confirmation.

Voice

  • Clear
    Plain words, one idea per sentence.
    Write "Use these tools together to get more done"
    Not "Leverage the platform to operationalize cross-functional synergies"

  • Professional
    Competent, not stiff or silly.
    Write "The forecast didn't finish. Here's why..."
    Not "Oopsie! Something went sideways."

  • Clear and decisive
    State things, don't hedge.
    Write "This report covers last quarter's demand"
    Not "We think this may contain the data you might want."

  • Human
    Write like a person; use contractions.
    Write "You're in"
    Not "User authentication completed successfully"

  • Concise
    Cut every word that isn't working.
    Write "Pick an option to start"
    Not "In order to proceed, please first select an option below."

Tone

  • Direct & Clear
    Fewer words, no warm-up.
    Use for routine actions, navigation, and system feedback.

  • Supportive & Friendly
    Warmth and reassurance.
    Use when users may need reassurance, guidance, or recovery.

  • Playful
    Light humor where it fits. Never in errors, legal alerts, or money/data moments.
    Use sparingly for low-risk moments and moments of delight.

Microcopy

  • Buttons
    Name the action.
    Write "Download report"
    Not "Click here"

  • Helper text
    Only when it prevents a mistake.

  • Confirmations
    Be specific. Explain risks.
    Write "Delete this scenario? You can't recover it."
    Not "Are you sure?"

  • Terminology
    One word per concept. A "scenario" stays a "scenario," not a "model."

Errors

  • Lead with what happened, then what to do next
  • Never blame the user
    Write "Use MM/DD/YYYY, like 03/15/2026"
    Not "Invalid date format"
  • Avoid unnecessary apologies
    Write "We couldn't load your dashboard. Refresh to try again."
    Not "Oops! Something went wrong."
  • Always give a way out
    Write "That filter returned nothing. Try a wider date range."
    Not "ERROR: Operation failed."
  • Match severity
    A wrong filter isn't a crisis. Lost data is.

Empty States

Never leave one blank; silence reads as broken.

  • First run
    Orient and invite.
    Write "No scenarios yet. Build your first one. [Create scenario]"
    Not "No items to display."

  • No results
    Explain and redirect.
    Write "Q5 forecast" not found. Try a broader search.
    Not "0 results found. Nothing matches."

  • Cleared
    Confirm and reassure.
    Write "Inbox zero. New alerts show up here."
    Not "List is empty."

AI-Generated Content

  • Be honest about what's AI.
    Don't hide that it's machine-made.
  • Calibrate confidence.
    Write "Projects ~14% growth; confidence moderate, last year's data had gap"
    Not "Sales will increase 14% in March"
  • Keep the human deciding.
    Write "Review 6 recommendations"
    Not "Approve all recommendations"
  • Make output easy to check, not just easy to accept.
    Show the inputs. A confident wrong answer is worse than an honest "I'm not sure."
  • Don't fake humanity.
    Write "Here's what I found"
    Not "I'm excited to help!"
  • When AI can't do something, say so and point elsewhere.
    Write "I can't access live pricing yet. Pull it from the Pricing dashboard."
    Not "I'm unable to process that request."

Do's and Don'ts

Do

  • Lead with the point
  • Use "you" and "we"
  • Use contractions (but note: contractions can be a challenge for translation/localization; when in doubt, clarity and understanding are more important than friendly tone)
  • Name the action on every button
  • Always say what's next
  • Flag when AI is unsure

Don't

  • Blame the user
  • Apologize theatrically
  • Hedge
  • Use jargon to sound smart
  • Pass off an AI guess as fact
  • Leave an empty state empty