A presentation that tells a story with data aims to make findings understandable in context and with their limitations, rather than simply listing numbers. First identify the question your audience is seeking to answer; then choose the evidence and narrative sequence that illuminate it. Give each slide a purpose: to pose the question, show evidence, or explain an interpretation. The steps below can help you shape a traceable narrative from your data.
1. Define the presentation’s scope and audience
Note the data source, scope, and limitations: What period is being examined? Which records are included or excluded? What are the unit of measurement, sample, and comparison group? If you cannot verify the source or reproduce the calculation, do not present the finding as a definitive result; state that it needs to be verified.
Describe the audience concretely, too: How familiar are they with the subject, what decision are they concerned with, and which terms might they not know? The Harvard Extension School course description for data storytelling notes that different types of data can be summarized and visualized in different ways, and that the course also covers narrative structure, audience analysis, and ethical data use (course description). Treat these not as a ready-made formula but as starting points for planning the presentation’s scope and level of detail.
2. Move from data to the main message
Before putting all your findings on slides, ask yourself: What changed or differed? Why does this matter to the audience? What evidence supports it? Write the answer in a single sentence. The sentence should express the finding shown by the data, not just the topic (“Monthly applications”): “Applications increased every quarter during the period examined.” Use this sentence only if your data truly supports it.
Choose data that helps the audience understand the main message or assess the relevant decision. Not every interesting number needs to appear in the presentation. However, if leaving out a comparison would change the meaning of the claim, explain the scope and your selection. Do not make the data fit a predetermined conclusion; do not present unsupported inferences as findings.
3. Build the narrative flow
For a short presentation, you could try this sequence:
- Question: What issue does the audience need to resolve?
- Context: What are the starting point, period, and measurement definition?
- Evidence: What is the main pattern visible in the data?
- Interpretation: What does the pattern suggest, and what does it not prove?
- Next step: What decision, additional analysis, or verification is needed?
This sequence can help prevent a run of charts whose connections are never explained. Penn LPS Online’s article on digital storytelling discusses framing for the audience, narrative structure, and how visual layout guides attention (read the article). The aim is not to create a dramatic story, but to show why a finding matters for the next slide.
4. Choose a visual that fits the finding
First identify the relationship you want to show. A line chart may be suitable for showing change over time, while a bar chart may work for comparing categories; these are options to consider based on your purpose, not rules that apply to every dataset. If you want to show a distribution, proportion, or relationship between two variables, another visual may be more informative. Check your draft: Do the axes, labels, and scale allow the data to be read accurately?
Check whether the title states the finding and whether the units and period are given on the axes. If you highlight the point you want to emphasize with color or a brief annotation, make the meaning of the colors clear. Since the choice of scale can affect perception, check whether it makes the difference appear larger or smaller than it is, and explain your choice if needed. If a slide has many series, labels, and numbers, reconsider whether the main comparison is still visible.
5. Example: a hypothetical service-usage presentation
The numbers below are invented solely to illustrate the method; they are not the results of actual research. Suppose a team is presenting monthly application counts to managers. Let the hypothetical quarterly values be 1,000 applications in quarter 1, 1,100 in quarter 2, 1,250 in quarter 3, and 1,300 in quarter 4.
- Slide 1 — Question: “Did the number of applications increase over the year?” The period and the definition of “application” are explained.
- Slide 2 — Evidence: A chart showing the four quarters is used. The title states the increase shown by the data; the axes and unit are explained.
- Slide 3 — Interpretation and limitation: It is noted that the increase in the numbers alone does not explain service quality or the reason for the increase. Whether the application channel or recording method may have changed is posed as a question to check.
- Slide 4 — Next step: A recommendation is made to examine the channel breakdown and past periods measured using the same definition. This is a step to verify, not a proven result.
The observation, interpretation, and next step in the example are kept distinct: it does not say, “The number increased, therefore it increased for this reason.” If you are describing methods and results, you can also read how to prepare a data science project presentation; if your focus is presenting analytical findings, see the guide to presenting data analysis results.
6. Review the risks of misleading storytelling
Presenting a relationship as a cause: Two measurements that change together do not, by themselves, prove causation. If the data and methods in your presentation do not support such an inference, describe the relationship and state that the cause is unknown.
Missing context: Do not give a percentage or total without the period, denominator, and definition. Check whether comparison groups were measured using the same method.
Selective data: Showing only the period that supports a claim may misrepresent the trend. Examine the entire relevant period and any contrary findings that could affect the interpretation.
Too many visuals: Consider removing a chart that does not contribute to the main question or moving it to supplementary material. Johns Hopkins Engineering Magazine’s article on data storytelling emphasizes designing visuals for the target audience and interpreting data ethically (read the article).
7. Final checklist and frequently asked questions
Before the presentation, compare the numbers with the source, recalculate totals and percentages, and review titles, axes, units, periods, and explanatory notes. Ask someone else to read the slide, ideally without seeing the source data, and explain the main message in their own words. This lets you check whether the message is clear from the presentation.
- Is the main message a statement the data truly supports?
- Are the source, period, definition, and limitations explained?
- Does each slide serve a specific question or finding?
- Does the visual fit the comparison you want to make?
- Are the scale, unit, title, and color explanations clear?
- Are interpretation and proven results kept distinct?
- Have the numbers and calculations been verified again against the source?
Should I put a chart on every slide?
No. Use a chart if it makes the message clearer. For a slide explaining context, method, or limitations, brief text may be more suitable.
What should I do if there is no strong conclusion in the data story?
Do not hide the uncertainty. Explain what was observed, which inference is not supported, and what additional data may be needed to answer the question. This also lets the audience see the limits of the available evidence.
← Back to blog
