I had a simple moment of clarity in a Vienna hotel room that changed how I evaluated my own readiness to perform.
I was working through a sequence I’d been drilling for weeks. It felt solid. I was hitting it consistently. And then, for no particular reason, I started narrating to myself out loud — describing what I was doing, as if explaining it to someone watching. Not the secret part. Just the visible action, the story around it.
Everything fell apart.
Not a little. Completely. The sequence that felt polished and automatic dissolved the moment I introduced the cognitive load of speaking. My hands didn’t know what to do without my full attention. The skill I thought I had turned out to be a skill I had only in silence.
I’d been training for the wrong conditions.
What Automation Actually Means
Automation is the nervous system’s way of compressing a complex sequence into a single unit. Instead of thinking about each component of a movement in sequence — step one, step two, step three — an automated movement runs as a single command. You intend the outcome, and the body produces it without requiring moment-to-moment conscious supervision.
This is what makes expert performance possible. A concert pianist can play a technically demanding passage while reading the next page. An experienced driver can navigate a familiar road while having a conversation. The skill has been delegated to a part of the brain that doesn’t compete with conscious thought for processing resources.
But automation doesn’t happen automatically with repetition. You can repeat a movement many times and still not have it automated. What matters isn’t just the number of reps — it’s whether the reps have been done consistently and correctly enough that the nervous system has had reason to compress them.
And here’s the diagnostic: if a skill is truly automated, you should be able to perform it while doing something else cognitively demanding. If you can’t, it isn’t automated. It’s still conscious. And conscious skills have a problem: they compete with everything else you’re trying to do at the same time.
Why This Matters for Performance
Magic is inherently a dual-task activity. You’re executing technical work with your hands while simultaneously managing the audience, speaking, maintaining the character of the story, reading the room, and staying present to whatever is happening in the moment.
If the technical work isn’t automated, all of those tasks compete for the same cognitive bandwidth. And something is going to lose. Either the technical execution degrades under the load of talking, or you go silent and mechanical while you focus on the execution.
I’ve seen both happen. I’ve done both myself.
The mechanical silence version is actually more common among technically proficient but not-yet-performance-ready performers. They can execute the skill, but only in conditions of focused attention. When they have to talk, the skill demands that attention back, so they go quiet. The audience doesn’t know exactly what’s wrong, but they feel the performer going somewhere else.
The talking version is what happens when you prioritize presentation over execution — you keep up the narrative but the technical quality drops.
Neither is what you want. What you want is both: fluid, natural conversation running in parallel with clean, automatic execution. That requires genuine automation, not just high-rep practice.
The Test Itself
The talk-and-do test is exactly what it sounds like. You practice the technical sequence while speaking continuously.
Not narrating what you’re doing. Actually having a conversation — or simulating one. What I use: I tell a story from my day out loud while performing the sequence. Or I count backward from one hundred by sevens. Or I describe the plot of a film I’ve recently seen.
The content doesn’t matter. What matters is that the verbal task is real — it requires actual cognitive resources. If you’re just moving your lips without engaging language, you’re not creating the interference you need to test.
If the technical sequence stays clean while you’re genuinely talking, it’s automated.
If it degrades — hesitations, errors, unintended pauses in the movement — it isn’t. Not yet.
What I found the first time I used this test honestly: a lot of skills I thought were ready weren’t. The gap between “practicing in silence” and “performing while speaking” was much larger than I expected. Skills that felt solid under focused attention turned out to need significant additional work before they could run in the background of a conversation.
This was uncomfortable to discover but extremely useful. It gave me a reliable criterion for evaluating actual readiness versus the impression of readiness.
The Levels of Automation
Once I started thinking in terms of the talk-and-do test, I realized there are multiple levels of automation, not just one.
Level one: you can execute the skill while talking, but the execution is slightly degraded. Small errors appear. Timing gets sloppy. The skill is partially automated but not fully.
Level two: you can execute the skill while talking without degradation, but the talking is slightly degraded. You lose your train of thought. You pause mid-sentence. The skill is stealing some resources from the verbal task.
Level three: both run cleanly. The skill executes at full quality and the conversation flows naturally. This is genuine automation.
Level four — and this is where performance becomes art rather than just competence — you can execute the skill while talking while also observing the audience and adjusting in real time. Three tracks running simultaneously with no interference.
Most of the work in building toward real performance is the movement from level two to level three. Level one is table stakes. Level three is the minimum for professional work. Level four is what you’re working toward over years.
Building Toward Automation
Knowing what automation looks like doesn’t automatically tell you how to build it. A few things I’ve found useful:
Start the verbal layer earlier than you think you should. Don’t wait until you’re satisfied with silent practice to introduce the talking component. Introduce it when you’re at maybe seventy percent accuracy in silence. The interference will degrade the execution initially, but working through that interference is part of the training. You’re teaching the nervous system to compress the skill faster by creating pressure to do so.
Use progressively demanding verbal tasks. Counting aloud is easier than telling a story. Telling a story is easier than having an actual two-way conversation. Move up the ladder as each level becomes manageable.
Practice the actual words you’ll say during the performance. Not random talking — the specific patter, the specific narrative, the actual lines. This creates the right kind of dual-track training. You’re not just automating the technical skill; you’re also building the combined routine as a single unit.
Practice with an actual person when possible. The unpredictability of a real conversational partner — questions, reactions, unexpected responses — creates more realistic conditions than monologue practice. I’ve used friends and family for this, sometimes telling them what I’m testing for and sometimes not.
What the Test Reveals About “Readiness”
The talk-and-do test is ultimately a more honest answer to the question “am I ready?” than any other measure I’ve found.
“Ready” doesn’t mean “I can do this.” It means “I can do this while also doing all the other things performance requires simultaneously.”
A useful analogy from my consulting work: the difference between knowing your material and being ready to present it is the difference between understanding a framework in your head and explaining it clearly to a skeptical room while also reading the room, adjusting your pace, answering unexpected questions, and keeping an eye on the clock. Those are related but not identical skills. You need the underlying knowledge AND the additional capacity to run everything else at the same time.
Magic performance works the same way. The technical skill is necessary but not sufficient. Sufficient means it runs on its own, in the background, while the rest of your mind is occupied with the work of being present with an audience.
Test it honestly. Practice in the conditions of performance, not just the conditions of the practice room. The gap between those two things is where preparation lives.
And when everything falls apart the first time you introduce talking — which it will — don’t be discouraged. You’ve just discovered exactly where the next stage of your work needs to happen. That’s not failure. That’s precision.