The founding claim of NLP is simple and, at the time, slightly arrogant: excellence can be studied, coded and transferred. Not talent. Not magic. The actual patterns that excellent people run, underneath what they say about what they do.
Richard Bandler, who co-founded the field, was blunt about the structure of his own creation: NLP is an attitude and a methodology, not the trail of techniques it leaves behind. The methodology is modelling. Every technique in the field, from anchoring to the Meta Model, is somebody's finished modelling project. This post is about the method underneath all of them.
Where modelling came from
The story starts in the early 1970s at the University of California, Santa Cruz. Bandler, a psychology student, and John Grinder, a linguist, wanted to answer one question: what do excellent therapists actually do that makes them excellent?
They studied three people. Fritz Perls, the founder of Gestalt therapy. Virginia Satir, the family therapist. And Milton Erickson, the psychiatrist and hypnotherapist. The crucial detail is that Bandler and Grinder were not interested in the theories these men believed. They were interested in the patterns: the language, the timing, the attention, the sequencing, the way each man structured a session. They coded what they observed, and the results became the early NLP books: The Structure of Magic (1975), based on Perls and Satir, and Patterns of the Hypnotic Techniques of Milton H. Erickson (1975).
Then came the test that matters. They tried to teach other people to produce the same results. The model worked, or it was thrown out and rebuilt. That test, works or it is wrong, is the whole ethos.
What modelling actually is
Modelling is not imitation. Imitation copies surface behaviour, the mannerisms and the tone. Modelling extracts the pattern underneath: the beliefs, the states, the language, the physiology, the sequence, the context. Two therapists can behave completely differently and be running the same model. Two people can sound identical and be running different ones.
The process has five steps:
- Choose an exemplar. Someone genuinely, consistently excellent at the skill you want.
- Gather data. Watch them work. Record them. Ask them what they do, and then distrust the answer: people's self-reports about their own expertise are famously inaccurate. The model is in the behaviour, not the interview.
- Code the pattern. Identify the beliefs, the states, the language and the sequence that actually produce the results.
- Transfer it. Teach the pattern to someone who cannot do the skill yet, and see if their results move.
- Test and iterate. If the transfer fails, the model is wrong. Refine it. The test of a model is whether it works, not whether it is true.
That last sentence matters. Modelling is pragmatism on purpose: it does not care if the pattern is theoretically elegant, only if it transfers.
Modelling beyond therapy
The method was never limited to therapists. Once the pattern is clear, it applies anywhere there is excellence to study: sales, sport, teaching, leadership, parenting, negotiation. Find someone who gets results, code what they do that others do not, and build a teachable version of it.
Two honest caveats belong here. Much of NLP's modelling history was informal: brilliant observation, minimal formal research. And the expertise literature in psychology does support one piece of the idea: experts hold a large amount of tacit knowledge, patterns they cannot articulate, which is exactly why observation beats interviewing. So the defensible claim is that tacit expert patterns exist and can be made explicit by close observation. The grander claims, that modelling can transfer any skill to anyone in days, overstate the record.
Modelling versus techniques
The distinction shapes how you should think about NLP training. A technique is a finished model made by someone else, in someone else's context. Useful, but secondhand. Modelling is the ability to build your own: to look at any excellence and extract what makes it work.
This is also why we quote Bandler's line about the trail of techniques. Most NLP training sells only the leftovers: the swish, the anchor, the reframe, taught as recipes. A recipe without a cook. The attitude and the methodology, the layers that produce the techniques, are almost never taught. Twenty-four weeks is long enough to teach them, and that is what the programme does.
IONI's modelling project
In the later months of the programme, every participant runs a real modelling project: pick one person who is genuinely excellent at something, code the pattern, transfer it, and document the whole process, including what failed on the first attempt and what changed.
The completed project goes into the portfolio. Not a certificate, which only proves attendance, but a documented model with the student's own name on it, which proves the skill. If you want the foundation before the project, our guide to what NLP is covers the three layers in full.
To see the programme where modelling is the spine rather than an elective, look at the Spirit of NLP programme, or start with the free webinar.