Data storytelling and presenting insights
Analysis only creates value when someone acts on it. Learn how to turn findings into a short story with a clear answer, one supporting chart and a recommendation, and how to present it to busy people.
Imagine you find that revenue at one store fell for three weeks. You send a spreadsheet with six tabs and 20 charts. The manager opens it, feels lost and closes it. Your analysis was correct, but nothing changed.
Data storytelling is the habit of giving people the answer first, then just enough evidence to believe it, then a clear next step. It is not about drama or decoration. It is about being understood.
Start with the decision, not the data
Before you open any tool, write down two things: who will read this, and what decision they might make because of it. A store owner deciding whether to hire another barista needs different evidence from a marketing lead deciding which offer to run.
A good question to ask the person who requested the work: "If the answer is X, what will you do? And if it is Y?" If both answers lead to the same action, the analysis may not be needed.
A simple structure that works
Answer, evidence, action
| Part | What to write | Example for Brew & Bean |
|---|---|---|
| Headline | The answer in one sentence, with a number | "Weekday afternoon revenue at HSR Layout is lower than the other two stores, mainly because of fewer orders, not smaller bills." |
| Evidence | 1 to 3 charts or numbers that prove it | A bar chart of orders by hour for each store; average order value side by side |
| So what | Why it matters, in business terms | "Orders per day is the lever; each extra order adds roughly the average order value in revenue." |
| Recommendation | One specific action and how to measure it | "Run a 4 pm to 6 pm combo for two weeks and compare orders per day." |
| Caveats | What you are unsure about | "Only six months of data; one store had a road closure in March." |
Put the headline first. Many readers will not go further, so make sure the first lines alone would still be useful. This is sometimes called the answer-first or pyramid approach.
Charts that tell the story
A chart should make one point, and the title should state it. Compare two titles for the same chart:
- "Monthly revenue by store" (a label, the reader must work out the point) - "HSR Layout trails the other stores in every month" (a message)
Then help the eye: highlight the bar or line that matters in a strong colour and leave the rest grey; remove gridlines, borders and legends you don't need; label the lines directly instead of using a legend; start bar charts at zero.
Before and after
| Cluttered version | Clear version |
|---|---|
| Title: "Revenue" | Title: "Revenue dipped 3 weeks in a row after the price change" |
| Eight colours for eight items | One highlight colour, the rest grey |
| Legend at the side | Labels on the lines themselves |
| Y-axis starts at 90,000 to exaggerate a small gap | Bar chart axis starts at 0 |
| A 3D pie with 9 slices | A sorted bar chart |
Misleading charts are mistakes, even by accident
A truncated bar-chart axis, a cherry-picked date range, or a dual-axis chart that makes two unrelated lines look related can all mislead. If you notice you are choosing a chart because it supports what you want to say, stop and show the fairer one. Your reputation as an analyst depends on being trusted.
Correlation is not cause
Two things moving together does not mean one caused the other. Cold brew sales and temperature may both rise in summer; that does not prove a discount on cold brew raised sales. Say "is associated with" unless you ran an experiment (see *A/B testing and experiments*). Also look for another explanation: a festival, a holiday, a new office nearby, or a change in how the data is recorded.
Presenting to people
A 5-minute presentation
- 1State the question in one sentence and why it matters to this audience.
- 2Give the answer right away: the headline and the recommendation.
- 3Show the one or two charts that prove it. Walk the audience through: say what the axes are, then point at the key bit.
- 4Mention the caveats briefly and honestly.
- 5End by repeating the recommendation and asking for a decision or next step.
- 6Keep the extra analysis in an appendix; use it only if someone asks.
Know your audience's vocabulary
Say "orders per day", not "transaction count per daily grain". Say "we are fairly sure", not "p is less than 0.05", unless the audience is technical. If you must use a term such as median, explain it in a few words once.
How this comes up in interviews
Analyst interviews often include a case such as "Sales dropped 10% last month. What do you do?" or a take-home task where you must present findings. Interviewers look for: a structured approach (clarify, break the metric into parts, check data quality, form hypotheses), a clear answer in plain language, a sensible recommendation and honest caveats. Practise saying your answer aloud in under two minutes.
I am a data analyst writing for [audience, for example the owner of a coffee shop]. They need to decide [decision].
My findings:
- [finding 1 with number]
- [finding 2 with number]
- [finding 3 with number]
Write: (1) a one-sentence headline with a number, (2) three short bullet points of evidence, (3) one recommendation with how we would measure whether it worked, (4) two caveats. Use plain English and do not add any numbers I did not give you.What to expect: Check that no new figures appeared and that the recommendation actually follows from the findings. AI tends to sound more certain than the data allows, so soften the wording where needed.
Try it: Rewrite a weak report
Here is a weak update: "Attached is the sales analysis. Revenue by month, store and category is in the tabs. Cold Brew did well. HSR Layout is lower. Let me know if you have questions."
Rewrite it as a five-line message to the owner using the answer, evidence, so-what, recommendation and caveat structure. Invent plausible numbers for practice, but label them as examples.
Before you send or present
- The first sentence is the answer, with a number.
- Each chart has a title that states its message.
- There are no more than three key charts.
- The recommendation is specific and has an owner or a way to measure it.
- Caveats and data limits are stated.
- A friend who doesn't know the project could explain it back to you.
Key takeaways
- Start from the decision the reader has to make.
- Answer first, then evidence, then recommendation, then caveats.
- One chart, one message; the title states the message.
- Use colour and labels to guide the eye, and keep axes honest.
- Correlation is not proof of cause; say so, and look for other explanations.
- Practise explaining your result simply, in under two minutes.
Check yourself
Finishing the quiz marks the lesson as done.
1. Which is the best opening line for a report to a busy manager?
2. Why should a bar chart's axis start at zero?
3. Ice cream sales and sunburn cases rise together in summer. What is the right conclusion?
4. Which chart title tells a story rather than labelling?
5. What should come at the end of a short analysis presentation?
Saved on this device. Create a free account to keep it on every device.
Practise this with Sam
A real mock interview, out loud, with an honest scorecard. Free to start.