Slides that support, not compete

For every slide, ask what it shows that you could not say in a sentence. There are only three good answers:

• A picture. Where the queue backs up, what the new report looks like, the screen the customer saw. Describing a picture in words takes a paragraph; showing it takes a glance.
• A number that needs to be seen to be believed. "Six days" said aloud is a fact. "Six days" drawn next to "under four hours" is a gap the room can see the size of.
• A comparison. Before and after, this team and that one, the average and the outlier. The eye compares two shapes faster than the ear compares two numbers said a sentence apart.

If the answer is "it shows the words I am saying", the slide is not a slide. It is your notes, put where the room can read them, and it will win the race to the room's attention every time.

Noor's second support is "the average that hid it". Her first draft (constructed) said it in three bullets: average reply time under four hours in one week of May; one customer in Lisbon waited six days for a refund reply in the same week; averages hide the tail. Read aloud, the three bullets take her a while, and the room has finished reading them long before she has.

The rewrite is one picture: a short bar labelled "under 4 h" and, under it, a long bar labelled "6 days". No sentence on the slide at all. Noor says the why over it: "Same week. The dashboard said we were fine. This customer did not think so." The slide carries the gap; she carries the meaning. Neither could do the other's job.

Building a deck as a story, one idea per visual, layout, type and chart choice are taught properly next door, in Visual Storytelling ↗ and AI PPT ↗. Use them to make the slide well. This page is about a narrower thing those courses cannot see from the desk: what the room's eyes and ears are doing at the same moment while you stand next to the slide and talk.

That is also why the test here is spoken, not visual. A beautiful slide can still compete with you if it carries the sentence you are about to say.

Yue, Bjork and Bjork, "Reducing verbal redundancy in multimedia learning: An undesired desirable difficulty?" (Journal of Educational Psychology, 2013). The material was a narrated, animated PowerPoint on the life cycle of a star. Their first experiment had 105 participants. What changed between conditions was the on-screen text that ran alongside the same narration.

Their summary of the pattern, in their words: "exact correspondence between on-screen text and narration generally impairs learning, brief labels within an animation can improve learning." Text that was abridged or slightly different from the narration did better than text that matched it word for word.

The finding that matters most for a speaker is the one in the title. Learners preferred the identical text. It felt supportive. It was the version that hurt. Your own sense that "having the words on screen helps people follow" is exactly the judgement this study says you cannot trust from the inside.

Scope, stated plainly: these were students taking a computer-paced multimedia lesson, with a recorded narrator, tested on what they learned. It is not a study of live talks, all-hands meetings or adult professionals in a room with a speaker they know. It also does not tell you how many words on a slide is too many, or rank every possible kind of slide.

What transfers is the mechanism, not the effect size: when the eye and the ear are handed the same words, one of the channels is wasted and the two can interfere. That is a reasonable thing to carry into a room, labelled as an inference from a lab lesson, not a measured fact about your all-hands.

In multimedia-learning research this family of findings is known as the redundancy principle, a term associated with Richard Mayer's work. His own statement of it was not read for this page, so it is named here and not quoted. If you read a summary that turns it into a precise rule ("never put any text on a slide"), go back to what the studies tested: narrated lessons, mostly computer-paced, mostly students.

• Ten slides. His reason: "a normal human being cannot comprehend more than ten concepts in a meeting." That is asserted, not measured, and the number ten is doing a lot of work.
• Twenty minutes. The basis is his experience of pitch meetings: the time a founder actually gets in front of investors. It is a fact about that room, not about yours.
• Thirty-point font. His reason is that a room will "read ahead of you because it can read faster than you can speak." He also offers an alternative: divide the age of the oldest person in the room by two and use that as your minimum font size.

None of the three cites a study. Each is a seasoned investor's rule of thumb for pitches, published as a blog post in 2005.

Two useful things, both mechanisms rather than numbers. First, few slides per idea: a talk with one throughline and three supports rarely needs more than a slide or two per support, which is where Noor ends up. Second, big type forces fewer words. You cannot fit a paragraph at thirty points, so the rule quietly removes the slide that competes with you. Notice that his reason for the font is the same mechanism as Part 1: the room reads faster than you speak.

What does not survive is the precision. There is no evidence here that ten is better than eight or twelve, or that twenty minutes is a limit for an all-hands. Use the rule as a prompt to cut, not as a standard to meet.

AI PPT ↗ treats 10/20/30 the same way: a widely repeated convention, asserted from experience, with no study behind it. If you meet it as a "proven rule" in a template or a training deck, the source is a single post by one investor, and you can say so politely.