Data Visualization for Presentations

A chart in a presentation has one job: make the data's message obvious without requiring the audience to analyze the chart themselves. This is different from a chart in a dashboard or a report, where the reader has time and motivation to study the data in detail. Presentation audiences are listening to a speaker while simultaneously looking at a screen. The chart needs to communicate its point in the two to four seconds before the speaker moves on.

Most charts in presentations fail this test. They're designed for accuracy and completeness rather than for communication. A perfectly accurate chart with 12 data series, dual axes, and a legend that requires 10 seconds to parse is a chart that confused the audience while the speaker moved past it.

Chart Type Selection

Choosing the wrong chart type is the most common data visualization mistake. Every chart type has a job — a specific kind of comparison or relationship it's designed to communicate. Using a pie chart to show change over time or a line chart to compare categorical values produces a chart that fights the audience's natural interpretation.

Bar charts: Comparing values across categories. The most versatile chart for presentations. Use when you want to show which category is largest, how categories compare to each other, or how a single category performs over time (where time is the category dimension). Horizontal bars work better than vertical when category labels are long.

Line charts: Showing change over time. Use when the trend is the point — whether something is growing, declining, or stable. Multiple lines work when you need to show two or three trends together; more than three lines on a single chart usually requires a separate chart per series or a small multiples approach.

Scatter plots: Showing relationships between two variables. Less common in business presentations because they require more cognitive work to interpret. Use when correlation or distribution is the actual point — when you want to show that two metrics move together or that data points cluster in a meaningful way.

Waterfall charts: Showing components of change. Essential for decomposing a metric change from one period to another: opening value + gains - losses = closing value. Revenue bridges, ARR bridges, and budget variance analyses all benefit from waterfall format.

Area charts: Showing volume over time, especially when multiple components stack to a total. Useful for showing how the composition of a total metric changes — revenue by product line, headcount by department — but harder to read precisely than bar charts.

Avoid unless necessary: Pie charts (harder to compare than bars), 3D charts (distort proportions), dual-axis charts (often hide the relationship rather than clarifying it), and bubble charts in presentations (require too much active interpretation from an audience that's also listening).

The One Message Rule

Every chart in a presentation should communicate exactly one message. Not two. Not "here's all our data." One message.

Before designing a chart, write the message in a complete sentence. "Revenue grew 40% year over year." "The enterprise segment has 3x higher NRR than SMB." "Q3 pipeline coverage dropped below 3x for the first time this year." This sentence becomes the chart title.

A chart that illustrates "Revenue grew 40% year over year" should be designed to make that 40% growth visually obvious — through chart type, axis scaling, and annotation. A chart that requires the audience to identify the 40% figure themselves is a chart that hasn't finished the communication job.

If you find yourself needing to communicate two messages, make two charts. Audiences follow two simple charts better than one complex chart.

Color System for Data

Color in data visualization should carry meaning, not decoration. A chart with six colors because it has six data series but no semantic meaning to the colors is harder to read than a chart with one primary color and one accent color highlighting the key data point.

Establish a two-color system:

Primary data color: the color used for the main series or the data points the audience should focus on. Should be your brand's primary color or a high-contrast color that reads clearly at projection scale.

Accent color: a second color used to highlight the specific data point or bar that's the focus of the message. Often a contrasting warm color (amber or orange) when the primary is cool (blue or teal).

Gray for context: All supporting series — comparison periods, benchmarks, industry averages, prior years — should be displayed in gray. Gray communicates "this is for context, not the point." The primary color communicates "this is what you should look at."

A bar chart showing quarterly ARR with the current quarter in blue and all prior quarters in gray is immediately readable — the current quarter stands out and the trend is visible in context. The same chart with all quarters in different shades of blue requires the audience to identify which quarter is current.

Avoid: Red and green as primary chart colors, because red-green colorblindness affects roughly 8% of men. Red and green are appropriate for explicit positive/negative signaling (a red indicator for a missed target, green for an achieved one) but not for distinguishing series in a multi-series chart.

Annotation Strategy

Annotation — adding text, lines, or shapes to a chart to explain the data — is often the difference between a chart that communicates its point and one that requires explanation from the speaker.

Direct labels: Place data labels directly on the bars, points, or lines rather than using a legend. A legend requires the reader to look at the legend, identify the color, find the color on the chart, and associate it with the data. A direct label eliminates three of those steps.

Callout annotations: For a chart where a specific data point is the focus — a quarter where performance was particularly notable, an inflection point in a trend — a callout box pointing to that data point with a brief explanation makes the chart self-explanatory. "Enterprise launched in Q2" on a line chart that shows a growth inflection in Q2 tells the story without requiring the speaker to narrate it.

Reference lines: A horizontal reference line showing a target, a prior period average, or an industry benchmark gives the audience a comparison point. "A horizontal line at 120% NRR — the benchmark for best-in-class SaaS retention — with your company's line above it communicates outperformance more clearly than stating the number.

Trend lines: For scatter plots or data with variability, a trend line (regression line or simple moving average) makes the direction clear without requiring the audience to visually fit a line to the data themselves.

Progressive Reveal

In live presentations, building charts progressively — revealing data points or series one at a time — can be more effective than presenting a complete chart all at once, for certain types of content.

When progressive reveal works:

Comparison builds: Start with a baseline, then reveal the comparison. "Here's where we were a year ago. [pause] Here's where we are today." The contrast has more impact than showing both simultaneously.

Sequential reveals in waterfall charts: Revealing each component of a waterfall one at a time, with a brief explanation of each, guides the audience through the analysis rather than presenting the conclusion all at once.

When progressive reveal doesn't work:

When the full chart's pattern is the point. A line chart showing five years of growth should be shown complete — the shape of the trend is the message, and revealing it point by point obscures the shape.

When the audience is reviewing a deck asynchronously. Progressive reveals are powerless in a PDF or a shared document. Design your most important charts to communicate without animation.

Common Data Slide Mistakes to Avoid

Too much data on one chart. If your chart has more than three or four series, it almost certainly needs to be split into multiple charts or redesigned as a small multiples layout. The most common reason for chart overload is that the analyst is attached to showing all the data rather than the message the data supports.

Starting axes at non-zero. A bar chart with a Y-axis starting at 80 can make a 3% difference look like a 50% difference. Bar charts should almost always start at zero. Line charts have more flexibility — starting at a non-zero value is appropriate when the trend within a range is the point and starting at zero would make the trend invisible — but should be labeled clearly.

Stacked bar charts with too many segments. A stacked bar chart with six or seven segments per bar produces colors that are indistinguishable and segments that are too thin to read labels. Three segments is the practical maximum for stacked bars in a presentation.

Ratio metrics without both components. A chart showing "win rate" without showing deal count can hide that win rate improved because volume declined (you're cherry-picking easier deals) rather than because close rates improved. Show the denominator alongside ratio metrics when it's relevant to the interpretation.

Charts without context for scale. "Impressions grew to 2.4 million" — is that good? A chart that shows the trend but provides no benchmark makes it impossible for the audience to evaluate the result. Include a reference line for the target, prior period, or industry average so the audience can interpret what they're seeing.

The test for any data slide is whether the audience, without hearing the speaker explain it, would arrive at the same conclusion the speaker intends. If the chart requires narration to make its point, it's not finished.