Week 15: Presenting Data-Driven Insights

From Chart to Presentation

  • Last week: refine visuals so the message is easier to see
  • This week: organize those visuals into a story someone can act on
  • A final presentation is not a report read aloud
  • It is a decision-focused argument

The Presentation Has a Job

Your audience should leave knowing:

  1. What problem or decision motivated the analysis
  2. What evidence matters most
  3. What your results imply about decisions or actions

Why This Skill Matters More Now

AI can generate charts, summaries, and even draft presentations

What it cannot do is:

  • decide what matters
  • persuade a specific audience
  • build trust around a recommendation

Your advantage is not producing analysis; it is ensuring the analysis informs decision-making

Story Structure

Stories Need Tension

Good data presentations usually start with an imbalance1:

What is happening now?

What should be happening instead?

The gap between those two states gives the audience a reason to care.

Example Tension: Water cuts

Drought has reduced water availability in the district by 15%.

Water cuts are coming, but the district hasn’t decided how to apply them.

Beginning, Middle, End

Beginning

Problem, audience, context, imbalance

Middle

Evidence, interpretation, implications

End

Recommendation or call to action

The Audience Is the Main Character

The story is not mainly about:

  • how hard the project was
  • every dataset you found
  • every model you tried
  • every graph you made

It is about the decision your audience needs to make.

Before building slides, answer:

  • Who needs to act?
  • What decision do they face?
  • What evidence would change their mind?

Example: Frame the Water-Cut Story

Who needs to act?

The irrigation district board deciding how to allocate a 15% water cut.

What decision or question do they face?

Should the district apply the cut uniformly to all farms or target the cut to preserve higher-value water use?

What evidence would change their mind?

Evidence that one policy preserves more crop value than the other, and that the difference is large enough to matter for the district.

How this changes the deck

  • lead with the policy choice
  • show the value comparison
  • end with a recommendation

AI can analyze but not own the decision

AI can tell you:

  • what the data suggest
  • what policy might maximize value

It cannot:

  • take responsibility for the outcome
  • understand stakeholder incentives
  • persuade a board, client, or public

That is your role

Slides Logic

Titles Carry the Argument

Slide titles should be assertions, not labels.

Weak Stronger
Results Drought severity is associated with lower corn yields
Forecast Demand is projected to exceed current capacity by July
Conclusion The district should prioritize targeted water cuts

If someone reads only your slide titles in order, they should understand the story.

One Slide, One Idea

Each slide should make one claim.

If you hear yourself saying: “And also…

you probably need another slide.

Avoid Two Failure Modes

The document pretending to be slides

  • too much text
  • too many full sentences
  • audience reads ahead
  • speaker becomes redundant

The mystery slide

  • too sparse
  • no context
  • no clear claim
  • useless without explanation

Eight Minutes Requires Choices

Your final presentation should include:

  • problem and decision context
  • brief data and approach
  • key results and interpretation
  • implications and recommendations
  • visuals that support the message

It should not walk through every technical detail.

For each slide, ask:

  • What claim does this slide make?
  • What evidence supports that claim?
  • What can move to backup, appendix, or Q&A?

Slides Support the Speaker

In a live presentation:

  • the slide shows the structure
  • the speaker supplies the explanation
  • transitions connect the ideas
  • visuals carry the key evidence

If the slide says everything, the audience stops listening.

Your Comparative Advantage in an AI World

Old value:

  • running the model
  • producing the chart

New value:

  • framing the decision
  • interpreting the result
  • persuading the audience

If AI can run the regression, your job is to explain why it matters

Timing

A good 8-minute presentation might have:

  • 1.5 minutes for the problem and decision context
  • 2 minutes for data and approach
  • 2 minutes for key results
  • 2 minutes for implications and recommendations

1.5 minutes per slide is a good rule of thumb, but it depends on the content.

Final Checklist

Before presenting, check:

  1. Is the audience clear?
  2. Is the decision or problem clear?
  3. Do the titles tell the story?
  4. Does each slide make one point?
  5. Do visuals support the message?
  6. Is the call to action unmistakable?

Next Week: Project Presentations

  • Each team will have 8 minutes to present their project.
  • 4 minutes for questions on logic, evidence, and implications.
  • Written report also due Wed May 6.

Additional tips for presentations:

  • Practice with group members.
  • Time your presentation to ensure it fits within the allotted time.
  • Know your analysis so you can answer question