Stakeholder Personas
Profiles of common business audiences. Use these to calibrate vocabulary, depth, and framing before writing or presenting analysis.
Persona 1: The Executive (C-Suite / VP)
Role: Sets strategy, approves budgets, makes irreversible decisions
Available attention: 2–5 minutes for a written output; 10 minutes for a presentation
Technical comfort: Comfortable with percentages, ratios, and trends; uncomfortable with statistical notation
What they want: Bottom line upfront, confidence level, recommended action, risk if wrong
What they don't want: Methodology, caveats that don't change the decision, multiple charts of the same thing
Writing style for this persona:
- Lead with the finding, not the process
- One key number per slide
- Use "$" and "%" rather than statistical scores
- Active voice: "Customers in X segment churn 40% less" not "A 40% reduction in churn was observed"
Vocabulary to avoid: p-value, confidence interval, regression, model accuracy, feature, pipeline
Persona 2: The Product Manager
Role: Defines product roadmap; runs experiments; owns user metrics
Available attention: 10–20 minutes for a written analysis
Technical comfort: Comfortable with A/B test concepts, funnel metrics, cohort analysis; variable on statistics
What they want: User behaviour insight, experiment results, segment breakdowns, actionable recommendations
What they don't want: Deep statistical proofs; database-level detail
Writing style for this persona:
- Metrics by user segment and lifecycle stage
- "Statistical significance" is OK; "p-value" less so
- Include a "what this means for the roadmap" paragraph
- Charts preferred over tables
Vocabulary bridge: "The test was significant" is fine; "p=0.03" replace with "we're 97% confident"
Persona 3: The Finance / Business Analyst
Role: Models business performance; owns P&L or cost centre
Available attention: 30–60 minutes for a full report
Technical comfort: High for numbers and formulas; low for ML/statistics terminology
What they want: Defensible numbers, clear assumptions, ability to stress-test the model
What they don't want: Black-box outputs; vague ranges without methodology
Writing style for this persona:
- Show the formula or calculation, not just the output
- Provide an assumption log they can adjust
- Include a sensitivity table
- Excel-friendly output formats where possible
Vocabulary bridge: "The model predicts" → "Based on the formula: [X] × [Y] = [result]"
Persona 4: The Operations Lead
Role: Manages day-to-day team or process performance
Available attention: 5 minutes; acts on dashboards and alerts
Technical comfort: Comfortable with operational KPIs; low comfort with analytical methodology
What they want: Clear signal on what to act on today; ranked lists; thresholds and alerts
What they don't want: Historical analysis without a present-day action; uncertainty ranges that prevent action
Writing style for this persona:
- "Here are the 10 accounts to contact today, ranked by risk"
- Traffic-light status (red/amber/green) over percentages
- Avoid preamble — get to the list or the action quickly
Vocabulary bridge: All technical terms → outcome + action
Persona 5: The Marketing Lead
Role: Owns acquisition, retention, and brand performance
Available attention: 15–30 minutes for a report
Technical comfort: Comfortable with conversion metrics, attribution, and campaign KPIs; variable on statistics
What they want: Which channels/segments perform, ROI by tactic, customer behaviour insights
What they don't want: SQL-level detail; extensive caveats about data quality
Writing style for this persona:
- Compare against benchmarks (last period, industry)
- Include chart with trend over time
- Frame in terms of campaign impact, not model performance
- Acknowledge attribution limitations briefly, then move on
Vocabulary bridge: "Lift" is fine; "ROAS" fine; "regression coefficient" → "the relationship between spend and revenue"