Design Slides, Documents, and Infographics People Can Understand

Choose the message, preserve the denominator, and check the artifact people will actually receive.

A slide says that 75 percent of attendees found a workshop welcoming. Its source says that nine of twelve survey respondents selected that answer. Eighteen people attended.

The percentage is calculated correctly for one group and attached to another. Good typography cannot repair that change in meaning.

AI can organize a presentation, suggest a layout, and draft labels. Effective design also requires evidence checking: what each number counts, which comparison matters, and what the audience should understand or decide.

Begin with the communication job, then choose the visual form. A simple table with accurate labels can be more useful than a striking infographic built around the wrong claim.

Start with a complete small dataset

This original dataset describes a fictional practice reading session. It is synthetic teaching material, not a report of an event that occurred.

All attendees in this packet were registrants; all survey respondents were attendees. The other three respondents did not select “welcoming,” but the packet does not specify their answers. Six attendees did not respond. No information about why people were absent or did not respond is supplied.

Those definitions are part of the evidence. Keep them with the numbers when transferring data into slides or a handout.

Decide what the audience needs to know

For the fictional review meeting, the communication job is to explain attendance, summarize the limited feedback, and identify what the organizer still needs to learn.

A suitable main message is: “Attendance and feedback provide useful signals, but the feedback describes respondents rather than everyone who attended.”

That message is less dramatic than “The event was a proven success,” but it fits the evidence. If the audience needs to decide whether to repeat the event, the packet also lacks costs, organizer capacity, and other relevant considerations. Identify those gaps rather than letting a satisfaction percentage decide everything.

Ask:

Turn this verified teaching dataset into a three-slide explanation. Give each slide one main message. Preserve counts, denominators, definitions, and missing information. Use a table or chart only where it clarifies the point. Do not infer motives, claim measured impact, or treat survey respondents as all attendees.

Match the form to the relationship

Use a table when readers need exact values and definitions. Use a chart when a visual comparison helps them see a pattern. Use a process diagram when order or dependency is the point.

For this dataset, a small count table is already effective. A bar chart could show the four counts, but its labels should make the nested groups clear. The bars do not represent four independent populations.

A pie chart of nine positive responses and nine “negative attendees” would be wrong. Six attendees did not respond, and the packet does not classify the other three responses as negative. The attractive binary chart would invent information.

If using bars, begin the quantitative axis at zero unless there is a clear, explicitly explained reason for a different design. Keep units, labels, and relevant context visible. Do not create a data chart with an image generator that may alter numbers or geometry.

Build a three-slide explanation

Here is a complete content structure, not a generated presentation file:

Slide 1: Eighteen of twenty-four registrants attended. Show 18 ÷ 24 = 75 percent. Include the count definitions and a note that the source is a synthetic practice dataset. This slide answers the attendance question.

Slide 2: Nine of twelve respondents selected “welcoming.” Show 9 ÷ 12 = 75 percent of respondents. Also show twelve responses from eighteen attendees and six nonrespondents. Do not label the result as 75 percent of all attendees.

Slide 3: The next review needs missing context. Ask about the experience of nonrespondents, reasons for nonattendance if appropriate to investigate, costs, and organizer capacity. Label these as proposed questions, not findings. The packet supports no claim that follow-up has occurred.

A spoken explanation can connect the slides. Avoid putting the entire narration on each slide. Provide a separate handout or notes where the audience needs more detail.

Check the arithmetic independently

The attendance rate among registrants is 18 ÷ 24 = 0.75, or 75 percent. The survey response rate among attendees is 12 ÷ 18, approximately 66.7 percent. The favorable response share among respondents is 9 ÷ 12 = 75 percent.

Nine favorable responses also equal 50 percent of the eighteen attendees. That is a count-based statement about recorded favorable responses among all attendees; it does not establish that the remaining half had an unfavorable experience.

A useful label includes both numerator and denominator: “9 of 12 respondents.” The percentage can follow. This makes the claim easier to inspect and avoids letting a large number float free of its group.

Choose a rounding convention and keep it consistent. Displaying approximately 66.7 percent is suitable here; adding several decimal places would imply unnecessary precision for a count of twelve people.

Design for access as part of the content

Use meaningful headings, readable labels, a sensible reading order, and text alternatives for information carried visually. Do not use color alone to distinguish essential categories. A direct label is often clearer than a legend that requires repeated eye movement.

For web text, WCAG 2.2’s minimum contrast criterion generally calls for at least 4.5:1 for normal text and 3:1 for large text, with defined exceptions. Large text is generally at least 18 point, or 14 point bold, under the criterion’s definition. These are useful design references; passing a contrast check alone does not establish that a slide deck or document is fully accessible. W3C: Contrast Minimum

Test contrast against the actual background, including images or gradients. A text box that looks clear on a designer’s monitor may become difficult when projected or reduced to a phone screen.

Provide editable or selectable text where practical. An infographic should not be the only way to obtain the underlying numbers and explanation.

Review the exported artifact

Open the actual file you intend to distribute. Check page breaks, cropped labels, substituted fonts, links, reading order, and whether important text is selectable or available in an accessible alternative.

A correct source slide can export incorrectly. A chart label may wrap, a footnote may disappear, or a citation link may no longer work. Inspect the delivered version, not only the editor view.

If publishing to a website or learning platform, review it in that environment. Captions, alt text, and document structure can be affected by the publishing process. Do not mark the artifact as tested until you have performed the relevant checks.

The three-slide structure in this article is a planning example. No presentation export or accessibility test has been performed for it.

Ask a reader to explain the message

A small comprehension review can be more revealing than asking whether the design looks good. Ask someone to identify who the 75 percent refers to and what remains unknown.

If they say “three quarters of everyone liked it,” the design needs revision. The misunderstanding might come from the headline, chart, or spoken explanation. Locate the source of confusion and change it.

Record the actual feedback if you conduct the review. One reader’s success is useful evidence for that task, not a guarantee that every audience member will understand the artifact.

For students: make the evidence inspectable

Keep the original dataset, calculations, and source references with your project. Follow the course rules for AI-generated layouts or wording. Be prepared to explain every major number without reading the slide aloud.

For a group presentation, assign someone to check data and someone else to inspect the final export. A polished theme should not hide uncertainty about who verified the claims.

Practice: create three slides or one page

Use the synthetic dataset to produce the three-slide explanation or a one-page handout. Include the exact groups, a source note, and the unresolved questions. Check the arithmetic, layout, contrast, and exported version. If possible, ask a reader to explain one percentage.

Completion check: Every quantitative claim has the right denominator and source. The design makes the main message easy to find without inventing missing responses or outcomes. Any claimed export or usability check refers to the actual delivered artifact you reviewed.