Data Analysis · Tutorial 10 of 10
Data Storytelling: presenting findings, not just numbers
By the FUTEAC Team · 22 November 2026 · 13 min read · Beginner
In short
Data storytelling turns numbers into a finding with an action. It has three parts, in order: the finding in one sentence, two or three numbers that prove it, and one thing to do next. This tutorial works through six weeks of sales, finds that bags drive the shop, and writes it up in three sentences a manager can act on.
- A finding is a sentence, not a number. "Sales were 16,420" is a number. "Bags drive this shop" is a finding, and it tells the reader what to do next.
- Every story has three parts, in order: finding, evidence, action. Skip the evidence and the finding is an opinion. Skip the action and the finding is trivia.
- Two or three numbers prove a finding. Ten numbers bury it. Pick the strongest evidence and leave the rest in the appendix or on the dashboard.
- The chart proves the finding, it does not replace it. A chart without a sentence next to it makes the reader do the analyst's job.
- A truncated axis can make a small change look like a collapse. Bar charts start at zero, or the chart is lying even when every number in it is true.
- This tutorial feeds the SQL & Data Visualization course (/course/sql-and-data-visualization) and the Data Storytelling (/skill/data-storytelling) skill.
Before you start
What data storytelling is
A manager does not need your numbers. They need your conclusion, and they need to trust it. Data storytelling is the work between the analysis and the decision: take the numbers, find what they mean, and say it in words a busy person can act on.
Think of it like a doctor reporting a test result. The doctor does not hand you a page of blood values and walk away. They say what the values mean, which two numbers matter, and what you should do next. The page of values is still there if you want it. But the story is what you act on.
A data story has three parts, always in this order: finding, evidence, action. The finding is one sentence about what changed. The evidence is two or three numbers that prove it. The action is one thing to do next, named plainly. If any part is missing, the story fails: a finding without evidence is an opinion, and a finding without an action is trivia.
The data: the same six weeks, read differently
This tutorial uses the same six weeks of shop sales as Building Dashboards (/tutorials/data/building-dashboards): bags, notebooks, and pens, 1 Oct to 11 Nov 2026, 16,420 rupees in total. If you built that dashboard, you already know the numbers. Here they are again, because this tutorial reads them differently. A dashboard asks "how are sales going?" A story asks "what should we do about it?"
| Week | Bags | Notebooks | Pens | Total |
|---|---|---|---|---|
| Week 1 | 1,200 | 800 | 450 | 2,450 |
| Week 2 | 1,350 | 820 | 470 | 2,640 |
| Week 3 | 1,100 | 790 | 430 | 2,320 |
| Week 4 | 1,600 | 900 | 500 | 3,000 |
| Week 5 | 1,450 | 860 | 480 | 2,790 |
| Week 6 | 1,750 | 950 | 520 | 3,220 |
| Total | 8,450 | 5,120 | 2,850 | 16,420 |
The dashboard told you the totals. The story starts where the dashboard stops: what do these numbers mean, and what should the shop do?
Step 1: find the finding
A finding is a change, a comparison, or a surprise. Totals on their own are none of those. "Sales were 16,420" is a fact, and facts do not tell anyone what to do. To find the finding, compare things.
Compare the categories first. Bags brought in 8,450 of 16,420 rupees, which is 51.5%. Notebooks brought in 5,120, or 31.2%. Pens brought in 2,850, or 17.4%. One category is more than half the shop. That is a comparison, and it is the start of a finding.
Now compare over time, because a share that is growing is a stronger finding than a share that is standing still. In Week 1, bags were 1,200 of 2,450, which is 49.0%. In Week 6, bags were 1,750 of 3,220, which is 54.3%. The bags share rose by more than five points in six weeks, while pens grew only 15.6% from Week 1 to Week 6 against bags at 45.8%. Bags are not just the biggest category. They are becoming more important every week.
Now check the surprise, because a good story also says what did not happen. Week 3 was the worst week, 2,320, and every category dipped at once. That is not a bags problem or a pens problem. It looks like a slow week for the whole shop, so the story does not blame a product for it.
Put the comparisons together and the finding writes itself:
> Bags drive this shop, and they matter more every week.
One sentence. It is a change (the share is rising), a comparison (bags against the other two categories), and it points at a decision (protect bag stock first). That is what a finding looks like. If your sentence does not point at a decision, keep comparing until it does.
Step 2: pick the evidence
A finding needs proof, but it needs less proof than you think. Two or three numbers, chosen well, beat ten numbers every time. The reader will remember two numbers. They will not remember ten.
For this finding, the evidence is:
- Bags are 8,450 of 16,420 rupees, or 51.5%. This proves "bags drive this shop". One number, one share, and the share matters more than the total, because it shows the weight of bags in the whole shop.
- The bags share rose from 49.0% in Week 1 to 54.3% in Week 6. This proves "they matter more every week". Two shares, six weeks apart, and the direction is the point.
- Bags grew 45.8% from Week 1 to Week 6, while pens grew 15.6%. This proves the rise is bags growing faster, not the other categories shrinking. One comparison, and it closes the obvious objection.
Notice what was left out. The weekly totals are on the dashboard, and the reader can check them there. The pens total, 2,850, is true but it does not prove anything about bags. The average weekly sales, 2,737, is a fine KPI and it is not evidence for this finding. Evidence is not every number you know. It is the numbers that would change a sceptic's mind.
A useful test: read your finding, then read each piece of evidence and ask "so what?" If the answer is a shrug, cut it.
Step 3: write it in three sentences
Three sentences, in order: finding, evidence, action. Here is the whole story for this shop:
> Bags drive this shop. They brought in 8,450 of 16,420 rupees (51.5%), and their share rose from 49.0% in Week 1 to 54.3% in Week 6, while pens grew only 15.6% over the same period. Order bags first when stock runs low, and check bag stock before the weekend, when the shop sells most.
Read it back and check the three parts. Sentence one is the finding. Sentence two is the evidence: two numbers, both from Step 2. Sentence three is the action: order bags first, and name the moment it matters, before the weekend. The action names a person, in effect: whoever orders stock. An action with no owner is a wish, so name the job, not just the task.
Compare that with the version that lists numbers without a finding:
> Total sales were 16,420 rupees. Bags were 8,450. Notebooks were 5,120. Pens were 2,850. Week 6 was the best week at 3,220. Week 3 was the worst at 2,320.
Five facts, zero decisions. The reader finishes it knowing the same things and doing nothing different. That is the difference between a report and a story, and it is the whole skill this tutorial teaches.
Step 4: choose the chart that proves it
One chart proves this story, and it is the bar chart from the dashboard: bags 8,450, notebooks 5,120, pens 2,850. It proves the finding in one glance, because the bags bar is nearly twice the notebooks bar. Put it next to the three sentences, not instead of them. The chart is the proof. The sentences are the story.
Do not add the weekly line chart to this story. It is a good chart, and it belongs on the dashboard, but it proves a different point: that sales are growing over time. This story is about bags, not about growth, so the line chart would split the reader's attention. One story, one chart. If a second chart proves the same finding better, use it instead. Never use both.
Title the chart with the finding, not with the data. "Sales by category" describes the chart. "Bags are more than half the shop" states the finding, and the chart proves it. A reader who only reads the title should still learn the point.
The wrong turn: the chart that made a dip look like a collapse
Here is the easiest way to ruin a true story: draw an honest chart dishonestly. Take pens, the smallest category. Weekly pens sales were 450, 470, 430, 500, 480, 520. Week 3 dipped by 40 pens, from 470 to 430. That is a small dip, and the story should say so.
Now draw that as a bar chart with the vertical axis starting at 400 instead of 0. The Week 3 bar shrinks to a stub, and the Week 4 bar towers over it. The numbers are all correct, and the chart lies anyway. Measured on the rendered charts, the Week 3 bar looks 86% as tall as the Week 4 bar when the axis starts at 0, which matches the real gap: 430 against 500. With the axis starting at 400, the same bar looks 30% as tall as Week 4. Same data, same 40-pen dip, three times the visual drama.
A reader who sees the second chart will conclude that pens collapsed in Week 3, and they will be wrong. The fix is simple and it is a rule, not a judgement call: bar charts start at zero. Always. If a chart needs a cut-off axis to show a difference, the difference is too small to chart, and the finding belongs in a sentence instead.
This is why the story comes before the chart. The writer who knows the finding is "a 40-pen dip" will notice when the chart screams "collapse". The writer who starts with the chart will believe their own picture.
Try it yourself
Write the story for this shop yourself, in three sentences, using the same data. 1. Pick a different finding from the same six weeks. One honest option: "Sales are growing, but unevenly." Prove it with the Week 1 total (2,450), the Week 6 total (3,220), and the Week 3 dip (2,320). What action follows from it? 2. Write your three sentences: finding, evidence, action. Read them out loud. If any sentence needs a fourth sentence to explain it, cut words until it does not. 3. Pick the one chart that proves your finding. If your finding is about growth, it is the line chart, not the bar chart. Title it with the finding. 4. Now write the worst version: five bare facts, no finding, like the "before" panel above. Read both versions to someone and ask which one tells them what to do. The answer is the test. If you cannot find a finding in your own numbers, compare more: this week against last week, this category against that one, this branch against the average. A finding is always a comparison. Totals alone never produce one.
Common mistakes
- Listing numbers instead of stating a finding. "Sales were 16,420" is a fact, and a report full of facts is a table, not a story. Start with the sentence that says what the numbers mean, then prove it.
- Ten numbers of evidence. Two or three strong numbers prove a finding. Ten numbers make the reader stop reading after the third, and the strongest evidence may be the one they never reach.
- No action at the end. A finding without an action leaves the reader informed and stuck. Name one thing to do next, and name who should do it.
- A truncated bar-chart axis. Starting a bar chart above zero makes small differences look huge. The numbers stay true and the picture lies. Bar charts start at zero, every time.
- Hiding the bad week. Week 3 was the worst week, and the story keeps it in. A story that only reports good news is not a story, it is a sales pitch, and readers learn to distrust it. Say what the bad week was, say what it was not, and move on.
Frequently asked questions
How long should a data story be?
Three sentences for a finding, one page for a full report. A manager will read three sentences. A manager will not read three pages of numbers hoping a finding appears on page three. If your story needs more than a page, it is two stories, or a story plus an appendix.
What if the data has no finding?
It always has one, but it may not be the one you wanted. "Nothing changed" is a finding too, and it is often the most useful one: it tells the reader to stop worrying and spend the effort elsewhere. The mistake is inventing drama when the numbers are flat. Report the flat result plainly, with the numbers that show it.
Should the story go on the dashboard?
No. The dashboard shows the numbers; the story sits next to it, in a short report or a message. The Data Analyst roadmap (/roadmap/data-analyst) pairs them as one project for exactly this reason: "An interactive dashboard with a written report". The dashboard is what the reader checks. The story is what the reader acts on.
Can I use this structure for a school or work report?
Yes, and it is the same structure at any length: finding, evidence, action, repeated once per finding. A five-page report is usually three findings, each with its own evidence and action, and an introduction that names all three up front. The shape does not change. Only the count of findings does.
What do I do when two findings point in opposite directions?
Report both, and say which one matters more. In this shop, bags are growing fast and pens are flat. Those are not opposite findings, they are one finding with a consequence: put effort into bags, not pens. When findings genuinely conflict, name the conflict plainly and recommend the action for the bigger number. Hiding one finding to make the story tidy is how reports mislead.
Follow the Data Analyst Roadmap
Turn raw data into decisions with spreadsheets, SQL, Python and dashboards. Free, step by step, at your pace.
View the Roadmap