Modeling Excellence — The Discipline That Gives You Other People's Superpowers

Ask most people what NLP is and they will list techniques. Anchoring. Reframing. The swish pattern. Perceptual positions.

Here is the uncomfortable truth nobody puts on the brochure. Every one of those techniques is downstream of a single discipline, and that discipline is the one most training skips. It is the reason NLP exists at all. It is modeling.

Richard Bandler was not a psychologist and John Grinder was a linguist, not a therapist. What they did starting in the early 1970s was watch three master therapists work, extract the structure of what those therapists did that produced results, and turn that structure into something other people could learn. Virginia Satir, Fritz Perls, Milton Erickson. They did not ask what these people believed. They asked what they actually did, moment to moment, and how they did it. That process, that act of extracting a strategy from someone who cannot articulate it and installing it in someone who can learn it, is modeling. It is the original skill. Everything else is a byproduct.

Most people can't tell you how they do what they're good at. They just do it. Modeling is the extraction of the thing they can't say.

I have spent three decades applying this discipline, and I will tell you the same thing Bandler kept repeating in a long-form interview released in June 2025 on his YouTube channel, "Richard Bandler Reveals All! The Mind Behind NLP." NLP was never meant to be a fixed bag of tricks. It was meant to be an attitude of curiosity applied to subjective experience, a way of asking how does this work, not what should I believe. The techniques are the residue of that attitude. When you lose the attitude and keep only the residue, you end up with a certificate and a set of moves you cannot adapt to a live human in front of you.

The first time I saw modeling work at scale, nobody in the room knew the word.

In 2001 I was invited by a national sports psychologist to partner on an NLP application for the Singapore Armed Forces shooting contingent. The position matters more than sports here. As a small country, Singapore needs its deterrent capability to be credible. You could not field a shooting team that was declining. The previous year they had dropped from second to fourth place at the ASEAN Armies' Rifles Meet, and they were up against regular armed forces shooters with years of practice. Our team was full-time national servicemen, barely eighteen years old, in uniform for two years and gone.

I was given access to video-record eight shooters from the 2000 cohort. Two M16 gunners. Two pistol shooters. Two GPMG gunners. Two carbine gunners. These were not NLP students. They had never heard of neuro-linguistic programming. But when I watched them work and listened to their inner dialogue, the same patterns that NLP formalises were running in the background, and the best of them were achieving serious results.

One carbine shooter was characteristically cool. Nothing fazed him. He practiced his physical movements while he slept, to the point where he would twitch in the dark. When it rained at the range and everyone else stopped, he closed his eyes and ran the scenario at full intensity in his mind. A top athlete does not stop training because the venue is unavailable. He took the venue with him.

One GPMG gunner carried the heaviest weapon on the team and was not the biggest man on it. Before a three-hundred-meter rundown he would scream and shout, hype himself and his partner up, and amplify his physiology until it changed his strength. He also took the other teams by surprise, because nobody expects a gunner to be screaming before the whistle. It was not aggression he was using. It was state amplification, dressed as attitude.

Neither of these men could have explained what they were doing. The carbine shooter would have told you he was just thinking. The GPMG gunner would have told you he was getting pumped up. That is exactly the point. Masters cannot articulate their own mastery. When I delivered my report to the 2002 team, I did not say "go figure out your own approach." I gave a step-by-step flow chart of what to do at each point in the competition, atomized down to the behavioural level, so an eighteen-year-old who had never modeled anything could install the strategy instead of reinventing it.

In 2002, the SAF shooting contingent took first place at the ASEAN Armies' Rifles Meet for the first time in its history.

I hit the same wall in a completely different arena, and modeling is what broke it.

From 1997 to 2000 I kept placing second or third in Toastmasters evaluation competitions. I was doing the thing everyone does when they are close to the top. I was trying harder at the same approach. I would practice more, memorise more, polish more. And I would keep coming second.

At some point I stopped and asked a modeling question instead of a willpower question. What do the winners actually do that I am not doing?

For roughly six months, every other day, I went to a Toastmasters club as an evaluator and a table-topic participant. Win or lose did not matter. I was there to collect data. I watched, by my count, more than four hundred different speakers. I jotted down their body language and their tonality. I built two databases. One was a catalog of every approach that was not a winning approach, every gesture that undercut the speaker, every tonal tic that lost the room. The second was a catalog of the crisp, powerful, dynamic gestures and tonalities that I wanted to install in myself.

That was the year I stopped trying to be better at my own strategy and started installing a better strategy that belonged to someone else.

At the District 51 Evaluation Championship I drew the worst lot. I was thirteenth of thirteen, the last speaker. The judging format meant the audience had sat through more than an hour of evaluations before I got up. Cognitive fatigue was baked into the room. A willpower approach would have told me to just be more energetic, to fight the fatigue. A modeling approach told me something else. Audiences remember the beginning of a sequence and the end. I was the end. The salience was already mine if I did not squander it.

I decided the fatigue was an advantage, not a threat. The twelve competitors before me would be formulaic, because that is what tired judges reward and what nervous speakers default to. I would not evaluate as someone telling the speaker what they missed, which reads as "you were incompetent." I would evaluate in a way that let the speaker feel they had missed it by the width of a hair, and that if they caught it, it would transform their presentation. Uplifting. Compassionate. Counterintuitive for a competition setting.

It won me the championship. Not because I tried harder than the other twelve. Because I had spent six months extracting the structure of what winning looked like, then installed it.

When you watch an expert, you are not watching one skill. You are watching a stack of micro-decisions, micro-adjustments, and physiological states that fire in a specific sequence. The problem is the expert cannot see the stack either. They experience it as a single fluid action. "I just knew the moment to close." "It felt right." That fluency is the enemy of transmission, because you cannot teach a thing you cannot name.

Modeling solves that by treating the expert as a source of data rather than a source of explanation. You do not ask them to tell you the answer. You watch, you record, you ask them to walk through a specific moment, and you break the fluency apart into its components. What did you see first? What did you pay attention to? What did you say to yourself? When did your body shift? What was the trigger, and what was the first physical response?

The output is a strategy. In NLP terms, a strategy is a reproducible sequence of representational activity, what you see, hear, and feel internally, plus the external action, that produces a result. The TOTE model I have written about, Test-Operate-Test-Exit, is the architecture underneath every strategy. Modeling is how you discover the TOTE that a particular master is running.

Bandler and Grinder's original formulation in The Structure of Magic I (1975) was that how someone talks is not a description of their experience, it is their experience reorganised through language. The deletions, distortions, and generalisations in a person's sentences reveal the structure of their model of the world. To model excellence, you listen at that level, beneath the content, at the structure.

Robert Dilts, who extended the early work, made this explicit. He modelled excellence across therapy, business, and creativity using a systematic method of behavioural modelling, breaking experts down to beliefs, capabilities, behaviours, environment, and identity. The Disney Strategy I wrote about earlier is one of his modelling outputs, extracted from how Walt Disney actually made decisions, not from what he said about it.

The key distinction, and the one that separates modeling from imitation: imitation copies the surface movements. Modeling extracts the underlying sequence and reinstalls it in a different body, a different context, a different personality. The GPMG gunner's screaming would be useless to a shooter whose strategy runs on stillness. The point was never the scream. The point was the amplification of physiology before a physical task, and that principle transfers; the specific expression of it does not.

Here is the business problem hiding inside this skill. Your top salesperson, your best operator, your highest-performing hire, they probably cannot explain why they are good. Ask them and you will get a shrug and a proverb. This is not modesty. It is the same fluency blinding the carbine shooter and the GPMG gunner.

That poses a real problem for a founder. If the strategy lives only inside one person who cannot articulate it, you do not own the capability; you rent it from a temperament. The moment that person leaves, or burns out, or hits a context change, the capability leaves with them. You recruit replacements and hope. You write job descriptions based on credentials, because you have no way to describe the actual strategy that produces the result.

Modeling is the extraction tool that turns a one-person capability into an institutional one. When you model your best performer, you produce a blueprint the rest of the team can learn from, not imitate. You can hire for the placeholder skills and train the strategy, which is a far more reliable pipeline than hoping to get lucky with another natural.

The power of this shows up most in coaching, which is why I treat modeling as the core of how I work. When a founder hands me a problem, my first job is not to prescribe a technique. It is to elicit the strategy running underneath the problem, and to elicit the strategy they already have that solves it elsewhere in their life. Most people already possess the resource they claim to lack; it is running in a different context and has never been extracted and reinstalled where it is needed.

You can run this protocol solo or with the person you are modeling. It works for a top performer, a mentor, or yourself in a context where you have succeeded.

Step one, choose a narrowly-defined peak. Do not model "success." Model one specific result: the exact moment a deal closes, a presentation lands, a negotiation turns. Narrow beats broad, because strategy extraction works on a single sequence you can replay, not a life you cannot.

Step two, capture the sequence, not the story. Get the person to walk through one concrete instance in real time. Ask about the first thing they noticed, the first internal thought, the first body shift. Record their words verbatim if you can. Experts describe their process in command terms, "the trigger was…" and "then I…", and those commands are the gold. Resist the urge to write down their explanation and skip the sequence.

Step three, separate the invariant from the ornament. Run the sequence past a second instance and a second person if you can. What stays the same under different contexts is the strategy. What changes is ornament. The GPMG gunner's scream was ornament; the physiology amplification was invariant.

Step four, code it as a reproducible TOTE. Write it out as a test-operate-test-exit loop. What is the trigger that signals "start"? What is the first operation? What evidence tells them they are on track? What tells them to exit? This is the flow-chart language I used for the shooting team, and it is what makes a strategy teachable to someone who never felt it naturally.

Step five, install it in a learner with a feedback loop. This is the step almost everyone skips, and it is where modeling either works or silently fails. The learner runs the new strategy in a real or high-fidelity practice context while someone watches and corrects. The calibration loop matters here: without real-time feedback, the learner rehearses a strategy they have only half-extracted and conclude modeling does not work. With it, the strategy becomes behaviour in weeks instead of a concept that never fires under pressure.