An institute that teaches NLP publishing a list of what NLP cannot do looks strange until you think about it. The field's biggest problem was never its critics. It was its own marketing. For thirty years, NLP was sold with promises it could not keep, and the backlash was earned. This post is the honest side of the ledger: the fair criticisms, what actually failed testing, what the evidence genuinely supports, and how we handle the whole mess.
The fair criticisms
Start with the evidence base. Core NLP claims have been tested since the early 1980s, and the testing has mostly not been kind. Reviews in the mid-1980s by psychologist Christopher Sharpley found little empirical support for the field's central ideas. In 2010, psychologist Tomasz Witkowski reviewed 63 studies published over 35 years and concluded that the research base for NLP's core assumptions was weak. Those two findings, thirty-five years apart, are the honest summary of where this field stands with mainstream psychology.
There is a more favourable data point, and it deserves a fair hearing. A 2015 meta-analysis of NLP-based psychotherapy, published in Psychiatria Danubina, pooled 12 studies and found a moderate overall effect. But the caveats are exactly the ones you would want to know: the studies were small, mostly observational rather than randomised, and the authors were affiliated with an NLP organisation. When you hear someone quote "a meta-analysis shows NLP works", that is the meta-analysis they mean, and it is weaker than the quote suggests. The evidence question is covered in full here.
What failed testing
Three famous NLP claims did not survive contact with research.
Eye-accessing cues. The model that eye movements reveal which sensory system a person is using, and by extension whether they are lying, was tested and did not hold up. Watch people's eyes if you like; attention and eye movement are interesting. Using them as a lie detector will fool you. The dedicated post on eye accessing cues goes through it.
Fixed representational systems. The idea that each person has one dominant sense, visual, auditory or kinesthetic, and that matching it unlocks communication, also failed. People are more flexible than the model assumed. Sensory language is worth noticing in conversation, and "one dominant system" is not.
The rapid cure. NLP's early marketing promised instant transformations: phobias cured in one session, personalities rebuilt in a weekend. A 2010 review of NLP for phobias, in Complementary Therapies in Clinical Practice, judged the method promising and time-efficient, while noting plainly that experimental research in the area was limited. "Promising and under-researched" is a very different claim from "proven".
What NLP cannot do, full stop
- It is not treatment for diagnosed medical or psychological conditions. If you have depression, anxiety disorder, OCD, trauma or any clinical diagnosis, see a licensed professional. NLP is not a substitute.
- It is not mind control. It does not make unwilling people do things. Influence requires presence, attention and consent.
- It does not guarantee outcomes. No training, NLP or otherwise, can promise results in relationships, sales or exams.
- It is not a licence to practise therapy. In India, "NLP practitioner" is not a regulated profession, and no certificate from any institute changes that. We have written about the certificate problem separately.
Where NLP genuinely helps
With the failures named, the defensible core remains. NLP is genuinely useful as communication and attention training: hearing the structure of what people say, noticing state and pacing in a room, questioning your own frames, deliberately building useful states, and modelling skills from people who do them well. These are trainable, transferable and, in coaching, therapy-adjacent work and leadership contexts, repeatedly useful. Concretely: a coach who can hear that a client's "I am just not good at this" is a generalization, not a fact, and who can ask the one question that dissolves it, is more useful than a coach with a hundred techniques and no ears. That is the skill, and it is learnable. The introductory guide describes the three layers, attitude, methodology and technique, and why most training sells only the third.
How IONI handles it
Three decisions follow from all of this. We teach the failed claims explicitly, in the programme, so students learn what collapsed and why, instead of inheriting dogma. We issue no certificates; graduates leave with a portfolio of logged hours, recorded sessions and a completed modelling project, which is harder to fake and more honest to show. And we do not sell certainty. Our programme is 24 weeks of supervised practice, because skills, unlike promises, need reps and feedback.
If you read this list and still want to learn the field, that is exactly the reader we want: one who wants the defensible version. If you read it and decide NLP is not for you, that is a fine outcome too, and we would rather you decide on accurate information. See the Spirit of NLP programme, or begin with the free webinar, where we will happily tell you the same uncomfortable truths in person.