Skip to main content

Audience Voice Guide

Locked audience taxonomy for dashboard-walkthrough-generator. Four profiles. The skill maps any author-declared audience to one of these before generating output. No ad-hoc profiles.

Executive​

Who. C-suite, VP level, board, senior stakeholder with 5-10 minutes.

Tone. Terse. Headline first. One number per sentence. Minimal jargon. No methodology detail unless asked. Confidence in the number, not in the chart construction.

Length target. 30-second elevator plus a 2-minute full walkthrough. Any longer and the audience disengages.

Script shape.

  • Open with the headline number.
  • Name the one chart that carries the headline.
  • Walk through at most 3 supporting sheets.
  • Close with a one-sentence "so what" that the author provides (not fabricated by the skill).

Anticipated follow-ups (what to prep for).

  • Strategic. "What should we do about this." Not methodological.
  • Trajectory. "Is this trending up or down." Pre-answer with the data point.
  • Comparison. "How does this compare to last quarter / to target / to peers." Pre-answer if the data supports it.
  • Confidence. "How sure are we." Pre-answer with the data source and refresh date.

Analyst​

Who. Fellow analyst, data engineer, data scientist, technical reviewer.

Tone. Technical. Named fields, aggregations, filters, and computation methods. Longer walkthrough tolerated. Comfortable with jargon the analyst audience shares.

Length target. 10-15 minutes total. Each sheet gets a methodology beat.

Script shape.

  • Open with the data source and refresh date.
  • For each sheet: fields on Rows / Columns, aggregation, filters, mark type, computation notes.
  • Call out any LOD, table calc, or cross-DS join that shapes the result.
  • Close with known limitations and open methodological questions.

Anticipated follow-ups.

  • Methodology. "How is this computed." Pre-answer with the calc text or aggregation.
  • Scope. "What does the extract cover." Pre-answer with the data source filter.
  • Edge cases. "What happens for X." Pre-answer if visible in the data.
  • Reproducibility. "Can I rebuild this." Pre-answer with the source table and key fields.

Operations​

Who. Line manager, ops lead, account manager, anyone whose follow-up is "what do I do next Monday."

Tone. Action-focused. What to do next, by when, by whom. Named entities (accounts, regions, products) over abstract summaries.

Length target. 5-8 minutes. Front-load the action.

Script shape.

  • Open with the top-N action items the data surfaces.
  • Each sheet: who is in the top-N, what to do about them.
  • Avoid trend analysis unless it drives a specific action.
  • Close with "here is who to call this week."

Anticipated follow-ups.

  • Tactical. "Which account should I call first." Pre-answer with the ranked list.
  • Attribution. "Why is X in the top-N." Pre-answer with the field that drove the rank.
  • Timeline. "When did this change." Pre-answer with the time dimension on the chart.
  • Ownership. "Who owns this." Pre-answer if the data has an owner field.

External​

Who. Customer, partner, external audience, cross-industry group.

Tone. Context-first. Define jargon. No internal acronyms. Assume the audience does not know the data source, the business, or the team.

Length target. 10 minutes. Context eats time; budget for it.

Script shape.

  • Open with a 1-minute context block: what business, what data, what time period.
  • For each sheet: define the measure and dimension before showing the chart.
  • Avoid internal benchmarks ("below target") unless target is defined.
  • Close with a plain-language summary of what the dashboard answers.

Anticipated follow-ups.

  • Clarifying. "What does ARR mean." Pre-answer with the definition.
  • Scope. "Is this all customers or just enterprise." Pre-answer with the filter.
  • Framing. "What should I take away from this." Author owns the takeaway; skill does not fabricate.
  • Comparison. "How does this compare to industry benchmarks." Author owns; skill does not fabricate benchmarks.

Mapping ad-hoc audiences to locked profiles​

The author may describe the audience in non-standard terms. Map before generating.

Author's phraseMapped profile
"my boss"Executive if VP+ title, Operations if direct manager
"my team"Analyst
"my VP"Executive
"the board"Executive
"a customer"External
"a partner"External
"cross-functional group"Analyst by default; ask if that is wrong
"mixed audience"Pick one and warn: "walking through at the lowest-context audience in the group"

State the mapping in the output so the author can correct.

Voice violations to catch​

  • Using analyst-level jargon in an executive script.
  • Using executive brevity in an external context (loses the audience).
  • Fabricating conclusions in any voice (HC-3 violation regardless of audience).
  • Mixing voices mid-script.