Good slide design comes down to one message per slide, stated as a sentence at the top, shown with a picture, chart or diagram, and stripped of anything that is only decoration. Your voice carries the detail; the slide carries the evidence. Then make the deck accessible, and give every borrowed image, chart and quotation its source.
This guide is published by EdCitation, whose tools look references up in the publisher's record and never write, rewrite or edit any part of a talk or its slides. Its PowerPoint tips come from Penn State's assertion-evidence pages and the studies behind them, Richard Mayer's summary of his research, Microsoft's accessibility help, the University of Washington, and four universities' study-skills pages, all read on 27 September 2026. For the references slide, EdCitation's free Cite a source builds each entry from the record.
Delivery has its own guides in this series: presenting without reading your slides, speaker notes and citing sources aloud.
What makes a slide help rather than hurt?
A slide helps when it shows the audience what matters and nothing else. That is the short form of three of Richard Mayer's multimedia design principles. Mayer (2023) gives, for each, how many of his lab's experiments supported it and the median effect size.
Coherence: cut what is merely interesting
People learn more deeply when extraneous material is left out: 18 of 19 tests, median effect 0.86. The classic culprit is the "seductive detail", a fascinating fact or dramatic clip that is not part of the explanation. On a slide, it is the stock photo of a lightbulb or the second chart that makes the same point.
Signalling: show what matters
People learn more deeply when cues show how the essential material is organised: headings, an outline, arrows, a highlight on the part of the diagram you are discussing. This held in 26 of 28 tests, median effect 0.70, and Mayer notes it works best used sparingly, with audiences new to a topic.
Redundancy: keep your script off the screen
On-screen text repeating the narration did not help in most tests (8 of 12), and the median effect was small, 0.10. The effect fades or reverses when the text is short, the pace slow, the audience working in a second language, or there are no pictures. A few key words beside a diagram are fine; paragraphs you read aloud are not.
The limits of this evidence
Mayer says most of the supporting studies were short laboratory lessons, often recorded explanations of how something works, such as lightning or a tyre pump. They were not built around a live seminar talk or an assessed student presentation.
What is the assertion-evidence approach?
The assertion-evidence approach, developed at Penn State's College of Engineering by Michael Alley and colleagues, replaces PowerPoint's default slide (a topic heading over a bullet list) with a full-sentence headline stating the slide's main claim, supported by visual evidence. Garner and Alley (2013) designed it to apply coherence, signalling and redundancy on every slide at once.
How to build an assertion-evidence slide
Penn State's one-page checklist (Alley, n.d.) gives the recipe. In our words:
- Write the headline as a sentence the audience should take away, left-aligned, two lines at most. "Trabeculae make a bone light but strong" is an assertion; "Bone structure" is only a topic.
- Support it with visual evidence, not bullets: a photograph, drawing, graph, film, or words and equations arranged visually.
- Use body words only when needed. The checklist's pace: the audience reads no more than 20 words a minute.
- Keep lists to two, three or four items.
- Set the type: a bold sans serif such as Calibri, 28 point headlines, 18 to 24 point body, and 12 to 14 point, not bold, for reference listings. No all capitals, italics or underlining.
Step 5 assumes slides carry references, in small type under the evidence they support.
What the studies found
Garner and Alley (2013) randomly assigned 110 undergraduate engineers at a US university to one of two eight-minute recorded presentations on how MRI detects tumours. The narration was identical; only the eleven slides differed. The default-style slides averaged 41.5 words, the assertion-evidence slides 21.2, each with a relevant graphic.
- Essays: the assertion-evidence group explained the process better (effect size 0.81) and made fewer misconceptions.
- Recall a week or more later (the paper gives both seven and ten days): higher for assertion-evidence (effect size 0.89).
- Multiple-choice facts straight after: no overall difference; the gap showed only on questions needing inference.
- Effort: the assertion-evidence group found the talk less demanding, yet rated its amount of text as less appropriate. They learned more from slides they liked less.
Garner and Alley (2016) then turned to presenters: of 120 engineering students building slides to teach MRI, the 59 told to use assertion-evidence understood the content better on a surprise test within 24 hours than the 61 who chose their own format.
What the studies did not find
Both studies used engineering students, one technical topic, and slides built to the criteria of the approach's developer, who is a co-author. At Kuwait University, Alzayed and Alzamel (2023) had 108 engineering students design eight slides on additive manufacturing: the assertion-evidence group reported lower workload, but motivation, confidence and understanding did not differ, and nobody delivered the slides. Wolfe et al. (2024), whose three pilot studies we read in abstract only, found that decorative diagram graphics such as SmartArt can suggest relationships that are not there, and that long quotations are better shortened or paraphrased than read out.
How many words per slide?
No study has set a number of words per slide. The figures in circulation are rules of thumb, and they disagree:
- Six, ever: Seth Godin's first rule in a 2007 blog post, citing no study (Godin, 2007).
- 6×6 (six lines of six words): on San José State University's quick guide, which says it is adapted from computing-magazine articles of 1997 to 2001. We could not trace who coined it.
- About 25: Norwich University of the Arts, citing a presentation-skills book by Van Emden and Becker.
- 33 at most: Butler University's "rule of 33", with no source given.
- About 21: the average on Garner and Alley's assertion-evidence slides, one condition in one experiment, not a rule.
The 10-20-30 rule
The 10-20-30 rule is Guy Kawasaki's, from a blog post of 30 December 2005 (he writes it 10/20/30): ten slides, twenty minutes, no font under thirty points (Kawasaki, 2005). He wrote it for pitches to investors. His reason for ten, that people cannot take in more than ten concepts in a meeting, cites no study, and his fallback font size is the oldest listener's age divided by two. A useful discipline for a short persuasive talk; your brief's limits come first.
What do universities advise about slide design?
University study-skills pages agree on big type, few words and relevant images, and split on bullets:
| University | Words | Bullets | Type | Also |
|---|---|---|---|---|
| York | One point per slide | None | Big, sans serif | A full-screen image with a short summary sentence |
| Norwich University of the Arts | About 25 | Use them | About 24 point, sans serif | Contrasting colours; copyright-cleared images |
| Butler | 33 at most | Four or five short ones | Not stated | Too much text and the audience reads instead of listening |
| San José State | Six lines of six words | Reveal one at a time | 24 point minimum, titles 35 to 45 | Credit sources at the end of the slide |
York's page gives Ned Potter's five golden rules, among them no bullets at all (University of York Library, 2026); Norwich (n.d.) and Butler (Butler University Libraries, 2026) keep bullets but cap them. Where your course sets a template, follow it. Otherwise the evidence above favours York: for explaining how something works, a sentence headline and a picture beat a list.
One caution: Norwich's page opens with a claim that we learn 90% of information visually. We could not find a study behind that figure, and none of the research above measured learning that way.
How do I make accessible slides?
Accessible slides work for everyone in the room and for anyone who opens the file later, including screen-reader users. Microsoft (n.d.-a) and the University of Washington (n.d.) agree on the essentials:
- A unique title on every slide, because screen readers navigate by titles. Microsoft shows how to move a title off the visible area; Washington suggests "Slide 1 of 3" for slides sharing a topic, and warns that big bold text is not necessarily coded as a title.
- Reading order. Screen readers read objects in the order they were added. Check the Reading Order pane; Washington notes the Selection pane lists that order from the bottom up.
- Alt text for every image and chart that carries meaning, a sentence or two on what it shows and why. Purely decorative images can be marked decorative.
- Contrast of at least 4.5:1 for normal text and 3:1 for large text, meaning 18 point or 14 point bold (World Wide Web Consortium, 2026). Microsoft notes its checker misses text in transparent boxes over a background.
- Never colour alone, and type of 18 point or more, as Microsoft advises.
- Captions on video; PowerPoint for Microsoft 365 can also caption your speech live, given a reliable internet connection (Microsoft, n.d.-b).
Run Review, then Check Accessibility, before you send the file.
Slide design principles at a glance
| Principle | What to do | Evidence or source |
|---|---|---|
| One message | A sentence headline stating the claim | Garner and Alley (2013); Alley (n.d.) |
| Evidence, not bullets | A photograph, diagram or chart that shows the claim | Garner and Alley (2013); York (2026) |
| Coherence | Cut decoration and seductive details | Mayer (2023): 18 of 19 tests |
| Signalling | Arrows, highlights, an outline | Mayer (2023): 26 of 28 tests |
| Redundancy | No script on screen; a few key words at most | Mayer (2023): 8 of 12 tests, small effect |
| Words per slide | As few as the slide needs | Rules of thumb only; no study set a number |
| Accessibility | Titles, reading order, alt text, contrast, captions | Microsoft (n.d.-a); University of Washington (n.d.) |
| Sources | Every image, chart and quotation credited; a references slide | Alley (n.d.); built with Cite a source |
Our slide checklist
Our own, built from the sources above. Run it on the finished deck.
- Could each headline be true or false? If it is a topic, rewrite it as a claim.
- Would the slide still make its point with the words covered? If not, find the diagram, chart or photograph that shows it.
- Is anything only decorative? Delete it.
- Could you read any slide aloud as your script? Move those words to your notes.
- Can the back row read the smallest text?
- Does every slide have a unique title, sensible reading order and alt text?
- Does every borrowed image, chart and quotation carry its source, and is each on the references slide? Build the entries with EdCitation's Cite a source and check the list with Verify references.
- Does the deck fit the brief? Check your paper reads the brief's instructions into a checklist, free; count the slides yourself.
Which EdCitation tool helps with a slide deck?
EdCitation takes care of what a deck borrows: its sources. Every word and slide of the talk stays yours, since EdCitation never writes, rewrites or edits any of it, and that is why its references can be trusted: each one is looked up, not composed.
Cite a source is the best tool for the references slide for that reason. It builds each entry from the publisher's record, from a DOI, a title, a PubMed ID, an ISBN or a web address, in APA 7, MLA 9, Chicago 18 author-date, Harvard (Cite Them Right), IEEE and Vancouver, and screens the source for retraction, free with no account. A chatbot asked for the same reference writes one from memory. On 27 September 2026 we gave it the DOI of the Wolfe et al. study above, 10.1177/00472816231169433. It read Crossref, found no retraction notice, and returned in APA 7:
Wolfe, J., Shanmugaraj, N., Reineke, J., Caton Peet, L., & Moreau, C. P. (2024). Advancing the knowledge base on effective presentation slide design: Three pilot studies. Journal of Technical Writing and Communication, 54(3), 235–264. https://doi.org/10.1177/00472816231169433
It set the title in the sentence case APA 7 asks for, so it matches our references below as it stands. For a chart or photograph from a web page, give it the page's address; the page's details can be corrected before the entry is built.
Verify references, also free, marks each entry verified, doubtful or not found, flags retractions, and never shows a source it could not check as not found. Find sources searches about 300 million published works for the study a slide's claim needs.
Quick questions
What is the best slide design for a presentation?
For explaining something, the best-tested design is assertion-evidence: a sentence headline stating the claim, supported by a picture or chart instead of bullets. In Garner and Alley's 2013 study it beat default bullet slides on understanding and later recall.
How many words should be on a slide?
No study sets a number. Rules of thumb run from six to 33; one study's assertion-evidence slides averaged about 21. Use as few as the slide needs.
What is the 10-20-30 rule?
Ten slides, twenty minutes, no font under thirty points: Guy Kawasaki's 2005 advice for investor pitches. It is a rule of thumb with no study behind it.
How do I make my slides accessible?
Give every slide a unique title, check the reading order, add alt text, keep contrast at 4.5:1 or more, never rely on colour alone, and caption video and speech.
Can EdCitation make my slides?
No. EdCitation never writes or edits any part of a talk or its slides. Cite a source builds the references slide from the publisher's record.
References
- Alley, M. (n.d.). Checklist for assertion–evidence slides [PDF]. College of Engineering, Penn State. https://www.writing.engr.psu.edu/AE_checklist.pdf
- Alzayed, M. A., & Alzamel, D. (2023). Assertion-evidence versus traditional PowerPoint: An investigation of the impact of slide structure on engineering students' cognitive load, motivation, and performance. Journal of Engineering Research, 11(1), 329–339. https://doi.org/10.36909/jer.16963
- Butler University Libraries. (2026, August 25). Using presentation software. Giving Presentations. https://libguides.butler.edu/c.php?g=1190378&p=8706339
- Garner, J. K., & Alley, M. P. (2013). How the design of presentation slides affects audience comprehension: A case for the assertion–evidence approach. International Journal of Engineering Education, 29(6), 1564–1579. https://writing.engr.psu.edu/ae_comprehension.pdf
- Garner, J. K., & Alley, M. P. (2016). Slide structure can influence the presenter's understanding of the presentation's content. International Journal of Engineering Education, 32(1A), 39–54. https://writing.engr.psu.edu/Garner_et_al_2016.pdf
- Godin, S. (2007, January 29). Really bad PowerPoint. Seth's Blog. https://seths.blog/2007/01/really_bad_powe/
- Kawasaki, G. (2005, December 30). The 10/20/30 rule of PowerPoint. https://guykawasaki.com/the_102030_rule/
- Mayer, R. E. (2023). Research-based principles for designing multimedia instruction. In C. E. Overson, C. M. Hakala, L. L. Kordonowy, & V. A. Benassi (Eds.), In their own words: What scholars and teachers want you to know about why and how to apply the science of learning in your academic setting (pp. 143–157). Division 2, American Psychological Association. https://teachpsych.org/ebooks/itow
- Microsoft. (n.d.-a). Make your PowerPoint presentations accessible to people with disabilities. Microsoft Support. https://support.microsoft.com/en-us/office/make-your-powerpoint-presentations-accessible-to-people-with-disabilities-6f7772b2-2f33-4bd2-8ca7-dae3b2b3ef25
- Microsoft. (n.d.-b). Present with real-time, automatic captions or subtitles in PowerPoint. Microsoft Support. https://support.microsoft.com/en-us/office/present-with-real-time-automatic-captions-or-subtitles-in-powerpoint-68d20e49-aec3-456a-939d-34a79e8ddd5f
- Norwich University of the Arts Library and Learning Support. (n.d.). Presentations. https://library.norwichuni.ac.uk/study-skills/presentations/
- San José State University Library. (2026, August 15). Presentations. Operations and Supply Chain Management. https://library.sjsu.edu/c.php?g=230203&p=8701778
- University of Washington Accessible Technology. (n.d.). Making slides accessible. https://www.washington.edu/accesstech/slides/
- University of York Library. (2026, September 3). Basic slide design. A practical guide to presentations. https://subjectguides.york.ac.uk/presentations/design
- Wolfe, J., Shanmugaraj, N., Reineke, J., Caton Peet, L., & Moreau, C. P. (2024). Advancing the knowledge base on effective presentation slide design: Three pilot studies. Journal of Technical Writing and Communication, 54(3), 235–264. https://doi.org/10.1177/00472816231169433
- World Wide Web Consortium. (2026, June 1). Understanding success criterion 1.4.3: Contrast (minimum). Understanding WCAG 2.2. https://www.w3.org/WAI/WCAG22/Understanding/contrast-minimum.html