Data storytelling is the work of turning a spreadsheet, a report or a set of survey results into something a reader can actually follow: a finding, a cause, and a next step, carried by the right chart instead of buried under one that only looks impressive. A number on its own rarely sticks. A number with a cause attached to it, shown in a chart chosen for the job rather than for familiarity, almost always does. The data storytelling examples below come from real public reports, and every one of them can be opened and checked.
What data storytelling actually is
The habit splits into two separate skills that usually get taught as one. The first is choosing and drawing a chart honestly: picking the type that matches what the data is for, comparison, change over time, distribution, a relationship between two things, and drawing it without a cut axis, a cherry-picked date range or a 3D tilt distorting what the numbers actually say. The second is structure: organising a finding around a setup that states what was normal, a tension that names what changed, and a resolution that says what should happen next, the same shape a story has always had, applied to a dataset instead of a plot.
Public reports and handbooks are a good place to watch both skills in practice, because their authors were solving the same problem long before “data storytelling” became a phrase anyone used deliberately. The National Patient Safety Agency’s 2008 risk matrix compresses a genuinely hard judgement, how serious is this risk, into a five-by-five grid a member of staff without risk-management training can read correctly in seconds. The American Forests measuring guidelines do something similar with a formula instead of a grid: circumference plus height plus a quarter of crown spread, a single number that turns three separate field measurements into one comparable score. Neither document calls itself a data storytelling guide, but both are doing exactly what one teaches.
A data storytelling guide in three steps
The practical version of this data storytelling guide compresses down to three moves, applied in order. Start with the sentence the data needs to prove, not the chart, since most bad charts are really a symptom of skipping this step: “region three grew fastest” or “churn fell after the onboarding change” already names its own chart type before a single pixel gets drawn. Then pick the chart built for that sentence’s purpose, covered in full in the which chart should I use guide, or answered directly by the chart chooser tool in a few clicks. Finally, check the chart for the honesty traps that undo good chart selection anyway: a truncated axis, a cherry-picked range, a legend sitting too far from the lines it explains.
That third step matters as much as the first two, and it is the one skipped most often under deadline pressure. A perfectly chosen chart type, drawn with a cut axis, still misleads a reader just as effectively as the wrong chart type would have. Reading reports critically depends on the same instinct in reverse: knowing what to check before trusting a chart someone else built, which is the subject of its own full guide to reading a report elsewhere on this site.
Data storytelling examples from real documents
Some of the clearest data storytelling examples come from documents built to be used under pressure, not admired on a shelf. The European Athletics Championships Statistics Handbook is compiled edition after edition by a Finnish statistics firm specifically so a journalist mid-broadcast can confirm, in seconds, whether a result just set inside the stadium beats a record set decades earlier by an athlete from a different generation entirely. That is not a chart at all; it is a sorted, indexed table, because the actual reader need was precise lookup, not a picture of a trend. Choosing a table over a chart, deliberately, is itself a data storytelling decision.
A different kind of example comes from Adobe’s 2015 consumer research, which reported that a majority of surveyed consumers abandon content that runs too long, and a similarly large share stop reading over poor design alone. The finding is itself an argument for the discipline this guide describes: a report or a slide that buries its point under six charts and no throughline is asking for exactly the kind of patience that research found in short supply, even a decade ago. Structure is not decoration; it is the difference between a finding that gets remembered and one that gets skimmed past.
Practising the skills, not just reading about them
Reading a chart-selection rule is not the same as building the instinct to apply it under a real deadline, which is why every lesson in the Academy pairs its guidance with something to try directly: a chart chooser tool that recommends a chart type from two or three questions, a game that hands over twelve real questions already paired with a chart and asks which pairings actually hold up, a slider that blurs three slide designs to show which one still reads its message from across a room. The Lab collects all of these short, ungated exercises in one place, each built to be finished in a few minutes without an account or a download.
None of this replaces judgement, and none of it claims data storytelling reduces to a formula that removes the need for it. What a genuinely useful data storytelling guide can do is name the recurring mistakes early enough that they get caught before a chart ships, and name the recurring structure clearly enough that a finding, once found, actually survives the trip from a spreadsheet to a reader’s memory.
Two tracks, built to be used separately or together
The Academy runs as two short tracks rather than one long course. The first covers charts and slides directly: picking the right chart type, spotting the handful of tricks that make an honest chart look dishonest or a dishonest one look fine, structuring a slide so a message survives contact with a real audience. The second covers reading and building reports: where a document’s actual argument usually sits, what an executive summary is supposed to do that an introduction isn’t, what a margin of error actually claims and what it doesn’t. Either track stands on its own, and a reader who only ever needs to fix one recurring problem, a report that gets skimmed past or a chart that keeps landing wrong, can start on the lesson that matches it directly rather than working through the whole sequence first.
The public documents used as examples throughout both tracks are chosen the same way: for what they show about a real decision under real constraints, not for polish. A risk matrix built by a health authority that closed over a decade ago and a measuring formula built for foresters comparing trees across different states have nothing obviously in common with a quarterly sales deck, and that is exactly the point. The underlying moves, compress a judgement into something scannable, state a formula precisely, choose a table over a chart when a table actually serves the reader better, repeat constantly across fields that never call what they’re doing “data storytelling” at all. Naming the pattern once tends to make it easier to recognise the next time it shows up somewhere unrelated.