Storytelling with data

0:00 Storytelling with data

Here's a chart. Twelve coloured lines, a legend with twelve entries, a title that says Q3 performance. Your audience squints, someone asks what the purple line is, and the meeting slides into a discussion of chart formatting. Sound familiar? Data doesn't speak for itself. People do. In this lecture, you'll learn to move from exploring data to explaining it, choose the right chart, declutter and focus attention, visualise honestly, and build a narrative around numbers. By the end, you'll transform one messy chart into a decision chart.

0:38 Why it matters

Why does this matter? Because decisions increasingly rely on data, and most people in a meeting won't do the analysis themselves. They depend on you to show them what matters. A clear chart with a clear message can turn a debate into a decision in minutes. A confusing or misleading one wastes time or, worse, drives the wrong decision. And with AI tools making charts in seconds, the scarce skill isn't making charts. It's choosing the right message and showing it honestly.

1:14 Explore vs explain

Here's the concept. Exploratory analysis is what you do to find the insight: slicing data, trying dozens of charts, following hunches. Explanatory analysis is what you show an audience: the one or two charts that prove your point. Many people present their exploration, all the charts they made. Your audience doesn't need your journey. Here's the key idea: before you make a chart for others, write one sentence. The audience needs to decide this, and the insight that matters is that. Then build the chart to prove that sentence.

1:53 The detective's wall

Here's an analogy. Think of a detective in a crime drama. During the investigation, there's a wall covered in photos, maps and string. That's exploration. But when the detective explains the case in the final scene, they don't walk everyone through the whole wall. They point to three pieces of evidence that prove who did it, in order. Your data story is the final scene. Show the evidence that proves the conclusion, and keep the wall in your appendix.

2:27 Choosing and decluttering

Choosing the right chart is simpler than people think. Change over time: a line chart. Comparing categories: a horizontal bar chart, sorted from biggest to smallest. Parts of a whole: a stacked bar or a simple percentage bar. Pie charts are hard to read beyond two or three slices. Relationship between two measures: a scatter plot. And one important number: just show it big, with context, like up from last quarter. Then declutter. Remove gridlines, borders and 3D. Label data directly instead of using a legend. Grey everything, then colour only what proves your point.

3:08 Honest visualisation

Honesty in visualisation is non-negotiable. Bar charts should start at zero, because the length of the bar is the message. Truncating the axis can make a two per cent change look like a doubling. Line charts can use a narrower range, but make it clear. Don't cherry-pick date ranges to hide an inconvenient trend. Show uncertainty where it matters. And be careful with averages that hide important differences between groups. If your chart could be accused of misleading, it will be. And your credibility is worth more than any single argument.

3:48 Worked example 1: Fatima's bakery (illustrative)

A simple worked example. Fatima runs a small online bakery in Islamabad. She wants to show her business partner that weekend orders are growing much faster than weekday orders. Her first chart is a table of fifty-two weeks. She writes her sentence: we should add a second baker on Saturdays, because weekend orders have grown much faster. She makes a line chart with two lines, weekend in her accent colour and weekday in grey, labels them directly, and titles it: weekend orders have doubled since January, weekday orders are flat. Her partner agrees in one conversation.

4:30 Worked example 2: Kareem in Riyadh (illustrative)

Now a realistic business scenario, illustrative. Kareem is an analyst at a regional airline based in Riyadh. Leadership is debating whether to add capacity on a route. The dashboard has thirty charts. Kareem builds a data story in three charts. First, load factor on the route over two years: consistently high. Second, lost bookings from sold-out flights, estimated from search data, labelled with its assumptions. Third, the illustrative revenue impact of adding one weekly flight, with a range, not a single number. The action titles tell the story. Leadership approves a trial for one season.

5:11 Numbers people can grasp

Numbers need translation. People struggle with large numbers, percentages of percentages and abstract units. So give comparisons. That's about one customer in every five. That's roughly the cost of two extra staff for a year. Round sensibly. Nobody needs twenty-seven point four three per cent in a presentation; about twenty-seven per cent is fine. And tell the story in three beats: here's where we were, here's what changed, here's what we should do. That's the SCQA structure from the last lesson, applied to data.

5:48 Watch me: messy data → decision chart

Watch me do it. I take an illustrative CSV of monthly repeat-order rates. I write my sentence first: we need to fix the loyalty programme before peak season, because repeat orders have fallen every month. In a spreadsheet, I insert a line chart, remove the gridlines, start the axis at zero, and label only January and June. Title: repeat-order rate has fallen every month since January. Then I ask an AI assistant with data analysis to describe the trend and show its calculations. It claims a seven-point fall. I check in the sheet: correct. I also ask it for caveats. It notes seasonality. I add that to my notes.

6:35 Measuring success

How do you measure success with data stories? Did the audience reach a decision? Did they ask about the implications, rather than what the chart means? Could someone repeat your headline after the meeting? And did anyone later find a flaw you should have spotted? Ask a colleague to look at your chart for five seconds, then tell you what it says. If they can't, simplify it. That five-second test is one of the most useful habits you can build.

7:10 Common mistakes

Common mistakes. Presenting exploration instead of explanation. No headline, just a label. Too many colours and categories. Pie charts with ten slices. Truncated axes on bar charts. False precision. Dashboards used as presentations. And trusting AI-generated numbers or charts without checking the calculation. AI analysis tools are useful, but they can misread columns or make arithmetic errors. You're accountable for every number on your slide.

7:38 Recap and try this now

Let's recap. Separate exploring from explaining. Write your one-sentence message first. Choose the simplest chart that proves it, declutter, and use colour to focus attention. Visualise honestly. Translate numbers into things people can grasp, and tell the story in three beats. Use AI for analysis, then verify every number. Your try this now: take one chart you've presented recently and remake it using the five steps in the lesson, then run the five-second test on a colleague. Next, we'll work on delivery and handling questions.