How to present data in a presentation: say the number, then say what it means

You've done the analysis. The numbers are right. You put them on a slide, walk the room through it, and at the end someone asks a question that shows they missed the point entirely. Not because they weren't listening. Because a chart with six series, a topic headline that says "Q3 Results", and a speaker reading out "so this one's up 14.7 percent and this one's down 3.2" gives a listener nothing to hold on to.

Most advice on how to present data stops at the slide: pick the right chart, cut the clutter, use one accent colour. That advice is correct, and this guide covers it. But it treats the audience as readers. In a room, or on a call, they're listeners first. They hear "fourteen point seven percent" once, at speaking speed, while also trying to read a chart and work out why you're showing it. A reader can reread. A listener can't.

So this guide has two halves. The first is the slide: which chart the eye reads most accurately, and how to make the headline carry the conclusion. The second is the part competitors skip: how to say numbers out loud, how to make a big number mean something, how to present a percentage or a risk so people actually understand it, and when a story beats a statistic. It's a spoke under how to give a presentation, which covers structure and delivery in general.

Here is the sequence:

1. Write the one sentence your data proves before you open a chart tool.
2. Pick the chart the eye reads most accurately: usually bars or a line.
3. Put that sentence in the slide headline, as a full sentence, not a topic label.
4. Grey out everything except the one value that makes the point.
5. Say each number once, rounded, in its own short sentence, with a pause after it.
6. Give every big number a comparison so the room knows whether it's big.
7. Turn percentages and probabilities into counts: "1 in 10", not "10 percent".
8. Pair the data with one concrete case, but don't bury the case under more numbers.
9. Walk every chart in four beats: what it shows, how to read it, the point, so what.
10. Rehearse out loud, on a recording, and check every number you say against the slide.

Why is presenting data harder than writing it down?

Presenting data is harder than writing a report because a listener gets each number once, at your pace, with no way to go back. In a document, a reader can stop at "14.7%", look at the table, compare it with last year, and carry on. In a talk, the number arrives, and two seconds later the next one replaces it.

That's a working-memory problem, and it's why spoken data needs to be thinner than written data. The usual estimate for how much a person can actively hold at once is about four chunks of information (Cowan, Behavioral and Brain Sciences, 2001), and a two-decimal percentage with a unit and a comparison already uses most of that. If you say three such numbers in a row, the first is gone by the time you finish the third.

The practical consequence is simple and a bit uncomfortable: a data presentation should contain fewer numbers than the analysis it came from. Most of the numbers that took you a week to compute belong in the appendix or the handout. The talk carries the two or three that support the conclusion, said slowly enough to stick. The rest is there for questions; how to handle questions after a presentation covers having those ready without putting them all on screen.

Start with the one sentence your data proves

The first step in how to present data is not a chart. It's a sentence. Write down, in plain words, what the numbers mean: "Support tickets doubled after the pricing change, and almost all of the increase is billing questions." That sentence is your talk. Everything on the slide either supports it or comes off the slide.

This sounds obvious, and nearly everyone skips it, because the analysis felt like the work. The result is the "here is everything I found" presentation: twelve charts, each accurate, in the order you computed them, with the conclusion arriving on slide eleven if at all. Audiences sit through that politely and leave without the point.

Three questions that get you to the sentence:

• What changed, compared with what? Data only means something against a baseline: last quarter, the target, a competitor, the plan. "Revenue was 4.2 million" is a number. "Revenue was 4.2 million, 600,000 under plan" is a finding.
• So what? What should this room do, decide, or believe differently because of this? If nothing, it might not belong in this talk.
• What's the one number that carries it? Usually one value, or one gap between two values, does most of the work. That's the number you will highlight on the slide and say out loud twice.

If you're presenting upward, this sentence also becomes your opening line, because senior audiences want the conclusion before the evidence; how to present to executives goes into that "answer first" structure in detail.

Which chart should you use to present data?

Use the chart that encodes your key comparison as position or length, because those are the visual cues people judge most accurately. In practice that means bar charts for comparing categories and line charts for change over time. Pie charts, bubbles, and colour shading ask the eye to judge angle, area, or hue, which people estimate much less precisely.

That ranking comes from one of the most cited experiments in data visualisation. In 1984, William Cleveland and Robert McGill at Bell Labs asked people to judge the relative size of values drawn in different ways, then measured how far off they were (Cleveland & McGill, Journal of the American Statistical Association, 1984). Position along a common scale produced the smallest errors, then length, then angle and slope, then area, with volume and colour saturation at the bottom. A 2010 replication with Mechanical Turk workers found the same order (Heer & Bostock, CHI 2010, 2010).

For a spoken presentation, the accuracy ranking matters even more than in a report, because the audience gets a few seconds per slide. A quick guide by job:

What you're showing	Use	Avoid
Comparing a few categories	Horizontal or vertical bar chart, sorted	Pie chart with more than three slices
Change over time	Line chart; bars if there are only a few periods	Area charts that stack many series
Part of a whole	A single stacked bar, or a bar chart of shares	3D pie, donut with labels outside
One headline number	The number itself, large, on a near-empty slide	A chart that contains only one data point
Relationship between two variables	Scatter plot with the trend line and one point labelled	A table of correlations
Exact values people will look up	A small table, in the handout	A dense table on screen, read aloud

Two cheap fixes catch most bad charts. First, flatten anything 3D: perspective makes the front bars look bigger than they are, which breaks exactly the length judgement bars are good at. Second, sort bars by value unless the categories have a natural order like months, so the eye reads the ranking without effort.

How do you make a data slide make its point?

Make the slide headline a full sentence that states the conclusion, then highlight the one value that proves it and grey out everything else. The headline tells the audience what to see; the highlight shows them where. Without both, people read the chart in their own order and often land on a different point than yours.

The headline change has the stronger evidence behind it. Penn State's Michael Alley and Joanna Garner tested it directly: about 110 engineering students heard the same recorded lecture on how MRI detects tumours, but half saw conventional slides with a topic headline and bullets, and half saw "assertion-evidence" slides, with a sentence headline stating the main message over a visual (Garner & Alley, International Journal of Engineering Education, 2013). The assertion-evidence group showed better comprehension, fewer misconceptions, and better recall of key ideas on a test a week later.

So the change on a data slide looks like this:

• Topic headline (weak): "Support tickets by category, Jan–Jun"
• Assertion headline (strong): "Billing questions drove almost all of the rise in tickets after the price change"

The chart underneath doesn't need to change at all. But now the audience reads the sentence, looks at the chart to check it, and finds the gold bar that confirms it. That's a three-second loop instead of a twenty-second puzzle.

The rest of the clean-up follows from the same idea. Label the bar directly instead of using a legend, so nobody has to look away from the data to decode it. Drop gridlines and most axis labels if a direct label carries the value. Use grey for context and one strong colour for the point. If you've also got text on the slide, keep it to the headline; a paragraph under a chart means the audience reads instead of listens, and you end up reading it to them. How to stop reading off your slides deals with that habit directly.

Build complex charts in steps

If a chart genuinely needs several series, reveal them one at a time. Show the axes and say what they are. Add the baseline series and say what's normal. Then add the series that breaks the pattern, in your highlight colour, and say what changed. Each click adds one idea, which keeps the audience's attention exactly where your voice is.

How do you say numbers out loud in a presentation?

Say each important number once, rounded, in a short sentence of its own, then pause and say what it means. "Tickets went up by four hundred. That's double." lands. "So we saw an increase of about 412 tickets, or 103.5 percent, which is mostly billing" does not, because the listener is still processing the first number when the second arrives.

This is the half of presenting data that slide guides leave out, and it's where most spoken data presentations lose people. Some rules that hold up in practice:

• Round for the ear. "About four hundred" or "just over a hundred percent" is easier to hold than "412" or "103.5". Put the exact figure on the slide for anyone who wants it. Your voice carries the rounded version.
• One number per sentence. If a sentence contains two numbers, split it. "Revenue was 4.2 million. The plan was 4.8." gives each figure its own moment.
• Lead with the comparison when it's the point. If the story is the gap, say the gap: "We're six hundred thousand under plan." Then, if needed, the two values behind it.
• Say the unit every time. "Twelve" means nothing. "Twelve days" or "twelve percent" does. Listeners can't glance at the axis label while you're talking.
• Pause after the number that matters. A one-to-two-second silence after the key figure is the spoken equivalent of bold type. It gives the room time to register it. How to use pauses when speaking explains why that silence reads as confidence rather than hesitation.
• Say the key number twice. Once when you reveal it, once at the end of the section or the talk. Repetition is not padding when the audience can't reread.
• Slow down on the figure itself. Speakers who know the data well tend to speed up on numbers, which is exactly backwards. The number is the newest information in the sentence; it deserves the slowest delivery.

Here's the difference in a real line. The slide shows a churn rate falling from 6.8 percent to 4.1 percent after an onboarding change.

Hard to follow: "So churn went from 6.8 to 4.1 percent, which is a 2.7-point decrease or about 40 percent relative, after we launched the new onboarding in March."

Easy to follow: "After we changed onboarding in March, churn fell from about seven percent to about four. (pause) That's roughly three in every hundred customers who used to leave, and now stay."

The second version uses the same data. It just orders it for a listener: the cause first, then the change in round numbers, then a pause, then what it means in people.

How do you make a big number mean something?

Give it a comparison the audience already understands: a ratio, a rank, or a smaller unit. "300 million firearms" is hard to grasp; "about one firearm for every person in the country" is not. Researchers call these comparisons perspectives, and adding them measurably improves how well people remember and judge numbers.

The best test of this comes from Microsoft Research. Pablo Barrio, Daniel Goldstein and Jake Hofman took numbers from real news articles and ran three randomised online experiments with more than 3,200 participants (Barrio, Goldstein & Hofman, CHI 2016, 2016). People who read a number with a perspective sentence attached remembered roughly half of the numbers they saw, against about a third without one. The perspective also beat simply repeating the number, and it helped people spot when a figure had been tampered with.

~1/2 vs ~1/3

Source: Barrio, Goldstein & Hofman, CHI 2016

One detail from the same study is worth knowing before you write your own. Not every perspective helped equally. "300 million firearms, about one per person" improved recall a lot: exact recall rose from roughly 40 percent to nearly 55 percent. "7.9 billion dollars, about $25 per person" barely helped. A good perspective connects to something the listener can picture or already knows. A bad one just swaps one abstract number for another.

Three patterns that usually work:

• Per person or per unit. "That's about two hours a week for every engineer." Turns an aggregate into something a single person can feel.
• A familiar reference. "That's roughly the population of Manchester." "That's about the size of our entire marketing budget." Pick a reference this specific room knows.
• Time. "At our current rate, that's eight months of growth." Many business audiences think in time more naturally than in counts.

This is the same move as using an analogy to explain an idea, applied to a number. How to explain complex ideas simply covers the wider toolkit for audiences who don't share your expertise.

How should you present percentages and risk?

Present percentages and probabilities as counts of people or things wherever you can: "1 in 10 customers", not "10 percent of customers"; "of every 1,000 users, 30 hit this bug", not "a 3 percent incidence". Counts, which researchers call natural frequencies, are far easier to reason with than percentages, especially when one percentage has to be combined with another.

The landmark study here is Gerd Gigerenzer and Ulrich Hoffrage's work on how people reason about probabilities. They gave participants problems where they had to combine a base rate with a test's accuracy, the classic "if the test is positive, how likely is it that you actually have the condition?" When the information was given as percentages, 16 percent of answers were correct. When the same information was given as natural frequencies ("10 out of every 1,000 people have it; of those 10, 8 test positive..."), 46 percent were (Gigerenzer & Hoffrage, Psychological Review, 1995).

16% → 46%

Source: Gigerenzer & Hoffrage, Psychological Review, 1995

You won't often present a Bayesian puzzle at work, but the same thing happens every time you say something like "conversion is up 20 percent on a 3 percent base." Half the room hears "20 percent" and pictures a big jump. In counts, it's "we used to convert 3 in every 100 visitors, now it's between 3 and 4." Less exciting, and much harder to misread.

A few habits:

• Say "percentage points" when you mean points. "Up from 4 to 6 percent" is a two-point rise and a fifty-percent relative rise. Say which one you mean, or say both values and let the audience see the gap.
• Keep the denominator constant. "1 in 20 here, 1 in 50 there" forces mental division. "5 in every 100 here, 2 in every 100 there" doesn't.
• Show a count as a picture when it matters. A grid of 100 small squares with 5 highlighted makes "5 percent" visible in a way no pie slice can.

Should you tell a story or show the statistics?

Use both, but in the right order and in the right proportions: one concrete case to make the room care, then the one or two numbers that show it's not an exception. Stories are more memorable and more moving than statistics, and adding a pile of statistics on top of a good story can actually weaken it.

The most quoted evidence for the first half comes from Chip Heath's classroom exercise at Stanford, described in Made to Stick: after students gave one-minute persuasive speeches, the audience remembered the stories far better than the statistics, even though speakers used many more statistics than stories (Heath & Heath, Made to Stick, 2007). It's a teaching exercise rather than a controlled study, so treat the exact percentages people quote with care. The direction, though, matches the controlled research.

The second half is the more surprising finding. Deborah Small, George Loewenstein and Paul Slovic gave 159 people $5 each and a chance to donate some of it to Save the Children (Small, Loewenstein & Slovic, Organizational Behavior and Human Decision Processes, 2007). One group read statistics about food shortages affecting millions of children in Africa. One read about a single seven-year-old girl, Rokia, with her photo. A third read about Rokia and saw the statistics.

The single child roughly doubled giving. But adding the statistics to her story dropped it to $1.43, much closer to the statistics-only group. The researchers' explanation is that the numbers pushed people into a calculating frame of mind, which dampened the feeling the story had created.

You're probably not raising money for charity, but the lesson transfers to any data presentation where you want people to act. Don't open with a wall of figures and then tack on an anecdote. Open with one specific case: one customer, one ticket, one afternoon lost. Then give the one number that proves it's typical, not the eight numbers that prove it in every segment. Storytelling in public speaking covers how to build that opening case so it's short and concrete.

How do you explain a chart in a presentation?

Walk every chart in four beats: what it shows, how to read it, what the point is, and what it means for the room. Most presenters jump straight to the third beat while the audience is still working out the first two, so they're reading axes while you're making your argument.

Here are the four beats, with a script for a line chart of weekly active users:

1. What it shows. "This is weekly active users over the last year."
2. How to read it. "Time runs left to right; the grey line is last year for comparison, and the gold line is this year."
3. The point. "Look at where the gold line breaks away, here in June. (point at it) That's the week we launched the mobile app."
4. So what. "Since then we've been about a third above last year, every single week. The app is now our main growth driver, which is why the roadmap I'm about to show puts it first."

Beats one and two take ten seconds and save the audience a minute of confusion. They feel redundant to you because you've looked at the chart a hundred times. Your audience is seeing it for the first time.

A couple of delivery points while you do this:

• Point at the thing you're talking about. Physically, with a hand or a laser, or with an on-screen highlight on a call. The eye follows the gesture. Body language for public speaking covers gesturing toward a screen without turning your back on the room.
• Turn back to the room for beat four. Beats one to three are about the chart; beat four is about the audience. Face them for it.
• Stop talking while a new chart appears. Give them two seconds to look before you speak. Talking over the reveal competes with it.

What are the most common mistakes when presenting data?

The most common mistakes are showing too many numbers, using a topic headline instead of a conclusion, choosing a chart the eye reads poorly, and saying numbers too fast and too precisely. Almost all of them come from presenting the analysis instead of the finding.

Mistake	What it looks like	Fix
Too many numbers	Every metric from the dashboard, one slide each	Pick the two or three that support your one sentence; appendix the rest
Topic headlines	"Q3 Revenue", "Survey Results"	A full-sentence conclusion: "Q3 revenue missed plan because of two late deals"
Hard-to-read charts	Pies with seven slices, 3D bars, dual axes	Sorted bars or a line; one axis; one highlight
Rainbow colours	Every series a different bright colour	Grey for context, one strong colour for the point
Reading the numbers off the slide	"This is 14.7, this is 12.3, this is..."	Say the one number that matters, rounded, then what it means
Too precise for the ear	"103.47 percent"	"Just over double" on the voice; exact figure on the slide
No comparison	"We had 4,000 sign-ups"	"4,000 sign-ups, against a target of 3,000"
Percentages without a base	"Up 50 percent!" (from 2 to 3)	Say both values, or say it as a count
Skipping the read-out	Jumping to the conclusion while people find the axes	Four beats: shows, how to read, point, so what

How should you rehearse a data presentation?

Rehearse out loud, on a recording, at least twice, and listen specifically to the moments when you say a number. Data presentations fail in ways you can't catch by reading your slides silently: you rush the key figure, say a two-decimal value nobody can hold, or forget to say what a chart is before explaining it.

When you play the recording back, check these:

• Every number you say matches the slide. Mismatches happen more than you'd think when you round on the fly, and a sharp audience member will catch it.
• The key number gets a pause. If the most important figure in your talk goes by at the same speed as everything else, slow it down.
• No sentence has two numbers in it. Split any that do.
• Every chart gets beats one and two. If you hear yourself saying "and as you can see" before saying what the axes are, add the read-out.
• You can state the one sentence without the slides. If you can't, the audience won't be able to either.

General rehearsal method is covered in how to practice public speaking. The data-specific part is simply to listen as a listener: pretend you've never seen the numbers, and ask whether each one, at the speed you said it, would have stuck.

Key takeaways

• How to present data starts with one sentence your numbers prove. Write it before building any chart, and cut every number that doesn't support it.
• Use bar and line charts for the comparison that matters. People read position and length far more accurately than angle, area or colour (Cleveland & McGill, 1984).
• Put the conclusion in the slide headline as a full sentence and highlight the one value that proves it. Assertion-evidence slides improved comprehension and week-later recall (Garner & Alley, 2013).
• Say each key number once, rounded, in its own sentence, with a pause after it and a comparison attached. Perspectives like "about one per person" raised recall from about a third of numbers to about half (Barrio, Goldstein & Hofman, 2016).
• Present risks and rates as counts. Correct reasoning rose from 16 to 46 percent when percentages were rewritten as natural frequencies (Gigerenzer & Hoffrage, 1995).
• Lead with one concrete case, then one number. Adding statistics to a story cut donations from $2.38 to $1.43 (Small, Loewenstein & Slovic, 2007).

Frequently asked questions

How do you present data in a presentation?

Start by writing the one sentence your data proves, then build the slide to make that sentence obvious. Use a bar or line chart, make the headline a full-sentence conclusion, highlight the one value that matters and grey out the rest. When you speak, walk the chart in four beats (what it shows, how to read it, the point, so what) and say each key number once, rounded, with a comparison.

How do you present numbers in a presentation so people remember them?

Use fewer numbers, round them for the ear, and give each important one its own short sentence followed by a pause. Attach a comparison that makes the size meaningful, such as "about one per person" or "roughly double last year". In a Microsoft Research study of over 3,200 people, numbers paired with that kind of perspective were recalled about half the time, against about a third without.

What is the best chart for presenting data?

For most presentations, a sorted bar chart for comparing categories and a line chart for change over time. Experiments on graphical perception found people judge position and length most accurately, and angle, area and colour least accurately, which is why pie charts and bubble charts are poor choices for any comparison your argument depends on. Use a large single number when there is only one value to show.

How do you explain a chart in a presentation?

Use four beats. Say what the chart shows ("weekly active users over the last year"), say how to read it ("grey is last year, gold is this year"), point to the thing that matters ("the gold line breaks away here in June"), then say what it means for the room. Pause for a second or two when the chart first appears so the audience can look before you speak.

What is data storytelling in presentations?

Data storytelling means organising numbers around a conclusion and a concrete case, instead of presenting them in the order you analysed them. In practice: open with one specific example that makes the problem real, give the one or two numbers that show it is typical, and end with what the audience should do. Research on charitable giving found that piling extra statistics onto a story can reduce its effect, so keep the numbers few.

Should you read numbers off the slide when presenting data?

No. Put exact figures on the slide for people who want them, and say a rounded version that a listener can hold, followed by what it means. Reading out every value on a chart turns a presentation into a dictation, and the audience stops listening by the third figure. Pick the one number that carries the point and say that one clearly.

Conclusion

Presenting data well is mostly subtraction: one sentence, one chart that the eye reads easily, one highlighted value, and a voice that gives each important number its own moment and its own comparison. The charts get the attention in most advice, but the audience hears your numbers before they read them. So rehearse the numbers out loud. Record a run of your next data slide, play it back as if you'd never seen the figures, and fix every place where a number went by too fast to keep.