How Much Text Should Be on a PowerPoint Slide?

There is no universal number of words that belongs on every PowerPoint slide. Use the least text that lets the audience identify the point, read the necessary evidence, and understand any qualification in the real delivery setting. Put the explanation you plan to speak in presenter notes, and move reference detail to an appendix or a separate leave-behind when the slide cannot do both jobs clearly.

Rules such as 6×6 or 7×7 can expose a crowded draft, but they are ceilings, not design specifications. A six-word chart title, a 35-word technical diagram, and a 70-word asynchronous instruction slide can each be appropriate. The question is not “How many words?” in isolation. It is “What must this audience see, and what can they hear or read elsewhere?”

Start with the delivery mode

Text density should follow how the presentation will be used. Decide this before trimming.

A live slide and a leave-behind page are different artifacts. Trying to make one file behave perfectly as both often produces a slide that is too dense in the room and too vague when forwarded.

If the audience will primarily read rather than watch, compare the handoff requirements in web presentations versus editable PowerPoint. The correct output format may solve the density problem more honestly than shrinking another paragraph.

Why a fixed word count fails

Word count does not measure visual or cognitive load accurately. Ten short labels around a diagram behave differently from a ten-word sentence. A table may contain many words but still support a precise lookup. One unbroken sentence can feel dense even when its count is low.

The fit also changes with:

• room size, screen size, and viewing distance;
• font family, size, weight, spacing, and contrast;
• the number of separate text boxes competing for attention;
• whether the audience already knows the terminology;
• the complexity of the chart, image, formula, or process beside the text;
• language and translation length;
• whether the presenter is speaking the same words at the same time.

Microsoft's current PowerPoint guidance says to choose type that can be read from a distance, avoid text smaller than 18 points, and use short sentences or bullets that do not wrap when possible. Its more general introductory guidance notes that type below 30 points may be difficult for an audience to see. These are useful warnings, not a reason to force every template or footnote to the same size. Test the actual room, screen, and file. See Microsoft's tips for creating and delivering an effective presentation and basic PowerPoint presentation guidance.

There is also a listening problem. In controlled technical-instruction experiments, presenting identical written and spoken explanations at the same time was less efficient than separating or removing one channel under the tested conditions. That does not prove that all on-screen text is harmful. It is a reason to avoid making an audience read a paragraph while the presenter recites the same paragraph. The original study is indexed by the U.S. National Library of Medicine as When Redundant On-Screen Text in Multimedia Technical Instruction Can Interfere With Learning.

Give every visible word a job

Instead of starting with a word limit, assign the text to five roles.

1. Takeaway

The title should tell the audience what the slide is about or, when the evidence supports it, what they should conclude. It should remain understandable when someone scans only the slide titles.

2. Evidence

Keep the labels, values, categories, dates, units, and comparisons required to read the visual correctly. Do not remove a denominator, reporting period, or definition merely to make the slide look cleaner.

3. Qualification

Show the caveat that materially changes interpretation: a limited sample, a preliminary result, an excluded region, or another boundary. If removing a phrase would make the claim misleading, it belongs on the slide or directly beside the relevant evidence.

4. Source

Keep a concise, readable source note when the audience needs to verify the evidence. The full citation can move to notes or an appendix only when the visible reference still identifies the source unambiguously.

5. Narration

Context, transitions, examples, and explanations usually belong in the presenter's voice. Microsoft's documentation describes the Notes pane as the place for supplemental information that does not appear on the slide, and Presenter View can show those notes to the speaker without showing them to the audience.

This role-based budget prevents a common mistake: cutting essential evidence while leaving generic commentary.

Run the three-pass audience test

Use this test on the exported slide at the size people will actually see. The five-second pass is a practical diagnostic, not a scientific threshold.

Pass 1: Glance

Show the slide for five seconds, then hide it. Ask a reviewer:

1. What was the point?
2. What evidence carried that point?
3. What should happen next?

If the reviewer reads the title but cannot find the evidence, strengthen the relationship between them. If several messages compete, split the slide or choose one primary conclusion.

Pass 2: Listen

Present the slide aloud. Mark any sentence the audience must read while you are saying something different. Then classify it:

• If it is essential evidence, pause long enough for reading or simplify the surrounding slide.
• If it repeats the narration, shorten it or move it to notes.
• If it is supporting detail, place it in an appendix or handout.

The goal is not zero text. The goal is a deliberate relationship between the visual and spoken channels.

Pass 3: Handoff

Open the deck without the presenter and ask whether the intended recipient can still identify the point, evidence, caveat, and source. If not, choose one of three honest fixes:

1. add a concise explanatory line;
2. distribute notes or a companion document;
3. create a separate reading version rather than overloading the live deck.

Rewrite a dense slide without deleting the truth

Consider this illustrative source paragraph. The bracketed fields are placeholders, not real results:

Onboarding completion changed from [BASELINE] to [CURRENT] after [PRODUCT CHANGE], while 30-day retention moved from [BASELINE] to [CURRENT]. The sample includes [N] new accounts from [PERIOD]. Regional variation and the absence of a controlled comparison limit the conclusion.

A weak edit converts it into three vague bullets: “Completion improved,” “Retention increased,” and “Results vary.” It is shorter, but it removes the values, period, population, and limitation that make the slide credible.

A stronger live-slide structure could be:

• Title: “Completion changed after [PRODUCT CHANGE]; causality is not established.”
• Visual: the verified baseline and current values, labeled with the period and unit.
• Callout: the verified retention change, presented as a separate observation.
• Qualification: “[N] new accounts; regional variation; no controlled comparison.”
• Source note: the exact internal report, table, or approved dataset reference.
• Presenter notes: why the change was introduced, what remains uncertain, and what decision is requested.

The rewrite is not successful because it has fewer words. It is successful because every visible text role has a reason to exist and no fact was silently discarded.

Use AI to compress by role, not by percentage

“Shorten this by 50%” tells the model how much to remove, not what must survive. A safer prompt names the audience, delivery mode, locked facts, and output roles.

Rewrite the selected slide for a [LIVE / WORKING / ASYNCHRONOUS]
presentation to [AUDIENCE].

The slide's one job:
[WHAT THE AUDIENCE MUST UNDERSTAND OR DECIDE]

Preserve exactly:
[NUMBERS, DATES, UNITS, NAMES, DEFINITIONS, QUOTATIONS, SOURCES]

Keep visible:
1. one accurate takeaway title;
2. the evidence required to understand the chart or diagram;
3. every qualification that changes the claim;
4. a concise source note.

Move explanatory context to speaker notes. Do not invent a causal link,
merge incompatible metrics, shrink text below the approved template minimum,
or remove a source. If the content cannot fit honestly, recommend a split and
name what belongs on each slide.

Compare the result against the source line by line. Check every value, unit, reporting period, named entity, and qualifier. The broader AI prompt library for editing PowerPoint includes additional patterns for shortening copy while protecting facts and layout.

Know when more text is the right choice

More text can be appropriate when the slide is meant to be read independently, teaches a procedure, defines technical terms, records a decision, or preserves wording that reviewers must inspect exactly. Even then, density needs structure.

Use:

• a descriptive title;
• short sections with clear labels;
• visible hierarchy between primary and supporting information;
• enough line spacing and contrast;
• a logical reading order;
• an appendix when reference material overwhelms the main sequence.

Do not treat accessibility as a reason to remove context. Microsoft's PowerPoint accessibility guidance recommends unique slide titles, readable fonts, sufficient contrast, white space, built-in layouts, and a checked reading order. A sparse slide with an unclear title or inaccessible visual is not an improvement.

Check the PowerPoint, not only the editor preview

Text fit can change after export because of font availability, font substitution, theme behavior, or different rendering environments. Reopen the PPTX in the PowerPoint version used for delivery and check:

• titles, body text, labels, and footnotes at full-slide size;
• unexpected wrapping, clipping, or automatic font reduction;
• whether notes contain the detail removed from live slides;
• reading order and Accessibility Checker findings;
• whether the slide still makes sense in Slide Sorter and Presenter View;
• whether a recipient can edit the required text and visual objects.

Use the complete presentation quality checklist before delivery, especially when AI has rewritten multiple slides.

Edit text density in Yrkie

Yrkie can create a presentation from prompts and source files or import an existing PPTX into its Copyist workflow. For an existing slide, define the audience and delivery mode, select the text that needs work, lock facts and qualifications, and ask for a role-based rewrite. Review the result in the editable workspace, keep supported elements under your control, and export a PPTX for the final PowerPoint check.

No AI editor should be assumed to preserve every font, animation, effect, or PowerPoint object without review. If the dense text belongs in narration rather than on the slide, the guide to generating a presentation script from PowerPoint explains how to develop and rehearse that spoken layer. Start with a new or existing presentation in Yrkie.

Frequently asked questions

Is the 6×6 or 7×7 rule correct?

It is a useful warning against crowded bullet slides, not a universal target. Do not add words to reach the rule or delete evidence merely to stay under it. Test the slide's delivery mode, text roles, readability, and handoff instead.

How many words should be on a live presentation slide?

There is no reliable universal number. Keep the takeaway, essential evidence, material qualification, and source visible. Move spoken explanation to notes, then test the slide at the actual display size while presenting aloud.

Should a slide contain full sentences?

Use a full sentence when it communicates a precise takeaway or qualification more clearly than a fragment. Avoid paragraphs that the presenter will read verbatim. For instructions or asynchronous decks, short explanatory sentences may be more useful than compressed bullets.

What should I do when important text will not fit?

Do not shrink it until it becomes unreadable. Decide whether the content belongs in notes, an appendix, a companion document, or a second slide. If each part supports a different audience question, split the slide.

How this article was prepared

This article was drafted with AI assistance from a reviewed editorial brief. The Yrkie team checked the method against current product behavior, Microsoft's primary PowerPoint documentation, and the cited original research. The bracketed example contains no customer or product metrics, and no universal word-count claim is presented as a tested fact.