The worst mistake I made in my first keynote that incorporated magic was treating the effects as the talk.
I had good effects. I had a room full of people who were, by all measurable indicators, engaged. I had reactions — genuine ones, not polite ones. And at the end of the sixty minutes, I sat in a taxi back to my hotel and realized that nobody, including me, could have stated clearly what the talk was about. The magic had been entertaining. The message had been absent.
I had performed a collection of effects with some words around them. I had not delivered a keynote.
The difference matters enormously, and it took that failure — and several more in various registers — to develop the framework I now use to decide whether an effect earns a place in a speaking context. Six questions. Every effect needs to pass all six before I’ll put it in a keynote.
Question One: Does It Serve the Message?
This one sounds obvious and is surprisingly easy to get wrong.
An effect serves the message when it demonstrates, embodies, or illuminates the central argument of the talk — not merely when it relates to the topic in the loosest possible sense. If the talk is about the gap between perception and reality, an effect that demonstrates the limits of visual attention serves the message. An effect that is simply astonishing in a way that could fit any context does not serve the message, even if it’s surrounded by words about perception.
The test is brutal and simple: could you remove the effect and still make the point? If yes, the effect might be entertainment, but it’s not argument. In a keynote context, you need it to be argument. Otherwise you’re doing a magic show with slides, not a keynote with magic.
Question Two: Can the Whole Room See It?
Close-up magic is the work I love most. It’s intimate, direct, responsive, and allows for the kind of connection that stage work doesn’t. It’s also completely wrong for most keynote contexts.
A ballroom with two hundred people cannot experience an effect designed for three. Not because the method changes, but because the experience is fundamentally about proximity — the way the magic happens right in front of you, in your hands, at arm’s length. When that effect gets projected on a screen or described verbally to the back rows, something essential is lost.
For keynote work, I only use effects that work at scale. The spectator at the back of the room has to have the same essential experience as the one in the front row. This rules out most of my favorite material. It’s a real sacrifice. But a keynote that leaves half the audience experiencing the talk second-hand is a failed keynote.
Question Three: Does It Advance the Emotional Arc?
A talk has an emotional shape — it moves through opening, development, and resolution, and the audience’s felt experience should change as it progresses. Every element in the talk either advances that arc or interrupts it.
An effect that produces the biggest reaction of the evening in the middle of the talk is not automatically a good thing. If the biggest emotional peak comes at the twenty-minute mark and the next forty minutes are a gradual descent from that moment, the arc is inverted. The audience is working backward from an emotional high that has already passed.
The question isn’t “is this effect powerful?” The question is “where does it fit in the emotional journey, and does it move that journey forward?”
Question Four: Is It Self-Contained?
Keynote effects need to set up quickly, deliver clearly, and close cleanly. They cannot require ten minutes of backstory, specialized vocabulary, or context that the audience won’t have unless they’ve been briefed beforehand.
The best keynote effects carry their own explanation in their structure. The audience understands what’s been proposed before it happens, they follow the action as it develops, and the reveal is legible without external annotation. They don’t need the speaker to explain afterward what just occurred. The effect does its own work.
Effects that require significant setup tend to fracture the attention that the rest of the talk depends on. By the time you’ve set up the effect, the thread of the argument is somewhere the audience can’t fully find their way back to.
Question Five: Does It Create Wonder, Not Just Surprise?
These are not the same thing, and the distinction matters for retention.
Surprise is short-lived. An unexpected reveal produces a sharp reaction and then fades quickly — often replaced by the audience working out what they missed, where the trick was, what the explanation is. Surprise produces puzzlement.
Wonder is different. Wonder is the feeling that something extraordinary has occurred without a diminishing appetite for explanation. The audience isn’t looking for the method because they’re still inside the experience. That quality is what makes magic useful in a persuasive context — it doesn’t just entertain, it opens a perceptual space that’s receptive to ideas.
An effect that produces “how did he do that?” has produced surprise. An effect that produces “I don’t understand how the world just did that” has produced wonder. In a keynote, you want the second one.
Question Six: Will They Remember It Tomorrow?
This is the final filter, and it’s the most practical.
A keynote lives or dies on what the audience takes out of the room. If the talk’s ideas are sound but the magic is forgotten by morning, the magic hasn’t served its function. It needs to serve as a mnemonic — an anchored memory that carries the idea it was connected to.
The effects I’ve found most durable in this sense share a quality: they’re deeply personal to the participant. Something happened to a specific person in that room — not to “the audience” but to one human being who is now at home telling their partner what happened to them. That specificity of experience is what makes the memory stick. A room full of people watched a show. One person had an experience. The second one travels further.
The Framework in Practice
Running an effect through all six questions typically rules out about two-thirds of the material I’m initially excited about for any given talk. This is frustrating, and I’ve learned to expect it. The effects that survive are the ones that were genuinely designed for the function, not the ones I love most.
The talk I mentioned at the beginning — the one that was actually a collection of effects with words around them — would not have passed this filter. Not one of the five effects I used that day would have passed question one. They served the entertainment. They did not serve the message.
The framework doesn’t guarantee a great talk. But it prevents a talk from being a show in disguise.
The material I cut hurts to cut. I keep it. It waits for the right context. A keynote effect is a keynote effect; a close-up effect is a close-up effect. The craft is knowing which is which.