Presenting Data Insights

Why Presentation is Part of the Analyst's Job

Many analysts believe their job ends when the analysis is complete. It does not. An insight that is not communicated clearly has no impact. The ability to present data findings to both technical and non-technical audiences is one of the most career-defining skills a data analyst can develop.

The goal of a data presentation is not to show your work. It is to change what your audience believes or decides.

Know Your Audience Before You Begin

The same analysis requires completely different presentations depending on who is in the room.

Before building a single slide, answer: "What decision do I want this audience to make or change after seeing this?"

Structure: The SCR Framework

Use the Situation-Complication-Resolution framework for data presentations.

Situation: What is the current state? (Establish shared context quickly.) "Our monthly churn rate is 4.2%, roughly in line with our historical average."

Complication: What has changed or what problem exists? "However, churn among users who signed up in the last 90 days is 11.8% -- nearly three times the baseline."

Resolution: What should be done? "Improving the onboarding experience for new users in weeks 1-4 is the highest-leverage intervention. We recommend a/b testing a guided setup flow starting next sprint."

This structure works because it gives executives all three pieces they need: context, urgency, and action.

The One-Minute Rule for Slides

Every slide must communicate its key message within one minute. If it cannot, the slide has too much content.

Headline your charts. Use the chart title to state the conclusion, not the description.

• Weak title: "Monthly active users by platform (Q1-Q2 2024)"
• Strong title: "Mobile MAU fell 31% in Q2 while desktop grew 8%"

The strong title means that even a distracted executive glancing at the slide for five seconds understands the point.

One chart, one message. Do not put two separate insights on one chart. It dilutes both.

Choosing What to Include

An analysis may produce 40 findings. A presentation should include 3-5. Prioritise based on:

1. Decision relevance: Does this directly inform the decision at hand?
2. Actionability: Can the business do something with this finding?
3. Significance: Is the effect large enough to matter in practice?
4. Confidence: Do you have enough data to be certain this is real?

Everything that does not clear this bar goes in an appendix, available if questions arise.

Handling Uncertainty Honestly

Resist the temptation to overstate certainty. Communicating limitations builds long-term credibility.

Weak: "This will increase revenue by $200,000."
Strong: "Based on our analysis, we estimate this could increase revenue by $150,000-$250,000
         over 6 months, assuming the retention improvement holds at similar scale."

Phrases that communicate appropriate confidence:

• "Our data suggests..." (directional signal, not certainty)
• "With 95% confidence, the effect is between X and Y"
• "This finding is based on 3 months of data -- the trend should be validated over a full quarter"
• "We cannot rule out that [confounding factor] explains part of this effect"

Live Presentation Techniques

Lead with the answer. In most business settings, bottom-line-up-front (BLUF) is preferred. Start with your main conclusion, then show the supporting evidence. Do not make executives sit through 15 minutes of methodology to get to the finding.

Anticipate questions. Before presenting, list every reasonable challenge:

• "How do you know it is not just seasonality?"
• "Is the sample size large enough?"
• "Does this hold for our enterprise customers too?"

Prepare backup slides in the appendix for each. If the question comes up, flip to the slide. This signals rigorous thinking.

Pause after key points. Give the audience time to absorb a chart before explaining it. Silence is not failure.

Manage the Q&A. For questions you cannot answer: "I do not have that breakdown today, but I will send you that analysis by end of week." This is far better than speculating.

Written Reports vs. Live Presentations

For high-stakes decisions, do both: a live presentation followed by a written summary within 24 hours.

The Pyramid Principle for Written Reports

Barbara Minto's Pyramid Principle structures written documents so that the most important point comes first, supported by evidence.

Main recommendation (top of pyramid)
├── Supporting finding 1
│   ├── Data point 1a
│   └── Data point 1b
├── Supporting finding 2
│   ├── Data point 2a
│   └── Data point 2b
└── Supporting finding 3
    ├── Data point 3a
    └── Data point 3b

Busy stakeholders read the top and stop if the conclusion is clear. Detail-oriented stakeholders read all the way down. The pyramid serves both.

Key Takeaways

• The goal of a data presentation is to change what your audience believes or decides -- not to show your work.
• Know your audience and tailor language, depth, and metrics to what they care about before building any slides.
• Use the SCR (Situation-Complication-Resolution) framework to structure findings clearly and drive toward action.
• Headline your charts with conclusions, not descriptions. Every slide must communicate its message within one minute.
• Communicate uncertainty honestly. Overstating confidence damages credibility; appropriate hedging builds it.

Practice Exercise

Exercise: Rewrite these weak slide titles as strong, conclusion-driven titles.

1. "Customer satisfaction scores by region (H1 2024)"
   Strong version: ___________________________________________

2. "Revenue breakdown by product line Q1-Q4"
   Strong version: ___________________________________________

3. "Support ticket volume trend"
   Strong version: ___________________________________________

Then write a 3-sentence SCR summary for this scenario:
- Situation: Your e-commerce platform processes 50,000 orders per month.
- Complication: Cart abandonment rate increased from 62% to 74% over the last 8 weeks,
  coinciding with the new checkout flow launched in March.
- Resolution: What would you recommend?