Journal · NLP · What it is
NLP Techniques Practitioners Use: Anchoring, Reframing, Meta Model
Six NLP techniques with a two-line example each: anchoring, reframing, submodalities and the swish, meta model, Milton model, calibration, plus the evidence.
Valérie Fabre, Founder & Director, Harmonika Institute · September 13, 2026 · 9 min read
Key takeaways
- Six techniques do most of the work in practice. Anchoring, reframing, submodality shifts (including the swish), the meta model, the Milton model, and calibration with rapport. The other 40-odd named patterns are variations.
- Anchoring is conditioning with a timer. Our NLP manual sets 4 criteria and a window: apply the stimulus at peak intensity and hold it 7 to 10 seconds. Then test it without warning.
- The meta model is 3 questions, not 30. What is left out, what is over-generalised, what is distorted. Ask them and vague statements become workable ones.
- The evidence for NLP as a system is weak. A 2012 systematic review found little support for health outcomes; the parts that work overlap with conditioning, exposure and cognitive reframing, which have evidence under other names.
- Scope is coaching and communication. Our 8-day program teaches these tools for goals, habits and conversations, never for any condition.
Ask ten certified NLP practitioners what they actually do in a session and you will hear the same six things, in different order. Not the 60 named patterns in the big books; six. A coach in Denver anchors a calm state before her client's board presentation. A practitioner in Seattle asks one meta-model question and a client's “I always fail” becomes “I missed two deadlines in March”. This article describes those six techniques with a two-line example of each, says when practitioners reach for them, and states plainly what the research supports.
What “certified” means in NLP and whether you need the Master level are covered elsewhere. This is the toolkit.
What are the NLP techniques practitioners actually use?
Six, grouped by what they act on: state (anchoring), meaning (reframing), the structure of a memory or image (submodalities), language in (meta model), language out (Milton model), and observation (calibration and rapport). Everything else in the NLP catalogue is built from these.
| Technique | Acts on | Two-line example | Practitioners reach for it when | Evidence |
|---|---|---|---|---|
| Anchoring | Emotional state | Client recalls a moment of full confidence; at the peak, presses thumb and forefinger together 7 to 10 s. Later, the gesture recalls the state. | A client needs a resourceful state on demand: talks, exams, difficult calls | Classical conditioning is well established; NLP-specific anchoring trials are few and small |
| Reframing | Meaning of an event | “I only got three replies.” “You got three people to answer a cold email in a week.” | A client's interpretation, not the facts, is the obstacle | Cognitive reframing has decades of support under other names; NLP's six-step version is untested |
| Submodalities / swish | Structure of an internal image or sound | The cigarette image is big, bright, close. Shrink and dim it; let a bright image of the desired self snap into its place, fast, five times. | A habit cue or a dreaded image keeps returning | Imagery rescripting has support; the swish itself has a handful of small studies |
| Meta model | The client's language | “Everyone thinks I'm incompetent.” “Everyone? Who, specifically?” | A statement is vague, absolute or assumes a cause | Structured questioning is a validated coaching skill; the model's specific taxonomy is not tested |
| Milton model | The practitioner's language | “You may already be noticing how much easier this feels than you expected.” | Directing attention gently; hypnotic and conversational change work | Suggestion effects are documented in hypnosis research; conversational use is not separately tested |
| Calibration and rapport | Observation and matching | Practitioner notes breathing and posture; matches pace and vocabulary before leading. | Always, from the first minute | Behavioural mirroring improves rapport in social psychology; eye-movement “cues” are not supported |
Anchoring: how does the manual teach it?
As a five-step procedure with four quality criteria. Our NLP manual defines anchoring as associating a desired emotional state with a specific stimulus, often a physical touch, so that an inner experience is linked to an external trigger and can be accessed at will. The steps: choose the state to anchor; identify a reference experience, real or imagined, in which it occurred; choose the stimulus; relive the experience until the state builds; and apply the anchor at the moment of maximum intensity. The manual's four criteria are that the person must be fully associated into the state; the state must be pure, not mixed with unwanted ones; the anchor must be unique, a stimulus that will not fire by accident in daily life; and the timing must hit peak intensity. The first-level manual adds the hold: 7 to 10 seconds at the peak.
Then you test. The manual's instruction is to fire the anchor without warning the client and watch whether the state returns; if not, it was set too early, too late or on a diluted state, and you rebuild it. Variants follow the same logic: stacking several experiences onto one anchor, chaining anchors from a stuck state to a desired one, collapsing a negative anchor with a stronger positive one. Strip the vocabulary and this is classical conditioning applied deliberately, which is why it works and why it fades if not refreshed.
From the training floor
The commonest anchoring failure our NLP instructors see is timing. Students fire the anchor when the client says “okay, I'm there”, which is after the peak. The fix is calibration: watch the face and the breathing, not the words, and set the anchor when the colour rises, a second before the client would report it. Cohorts of ten practise this in pairs for most of a day before it becomes reliable.
Reframing: context, content and the six-step version
Reframing changes what an event means without changing the event. Our manual introduces it through reformulation: when a client says “I went to see my mother-in-law yesterday”, the practitioner replies “You managed to go and see your mother-in-law yesterday”, a reformulation that, in the manual's words, includes a presupposition favourable to change, restating what was said with a slight shift of meaning that creates an opening. The manual sets reframing beside four other reformulation moves (echo, deliberate error, humour, provocation) and warns that the last two need a solid alliance.
Three named forms. Context reframing asks where a behaviour would be useful: stubbornness at work is persistence in rehabilitation. Content reframing asks what else the same facts could mean: a rejection becomes information. The six-step reframe, taught in our third module, assumes an unwanted behaviour has a positive intention, asks the “part” responsible for three alternatives that satisfy it, and checks for objections. The first two are cognitive restructuring under another name; the six-step version has no research behind it.
Submodalities and the swish
Submodalities are the qualities of an internal picture, sound or feeling: size, brightness, distance, colour, volume, location. Our curriculum describes the technique as learning “how to shift the structure of an experience”. The working idea is that a memory's emotional charge tracks its structure more than its content: a humiliating moment recalled large, bright and close feels worse than the same moment recalled small, dim and distant. Practitioners elicit the structure, change one quality at a time, and check what shifts.
The swish is the best-known application. The client holds the cue image for an unwanted habit, big and bright; a small, dark image of themselves as they want to be sits in a corner; on “swish”, the two exchange places fast. Repeated five to seven times, the cue loses its pull. It resembles imagery rescripting, which has research support; the swish itself has a few small trials with mixed results.
The meta model: three questions
The meta model is NLP's tool for recovering information that has been lost between experience and speech. Our manual describes it as a questioning technique that introduces precision to return to the original experience and avoid interpretations and imprecision, and it sorts every vague statement into three kinds. Omissions: something is missing (“I'm better now”: better than what, at what?). Generalisations: one or several experiences have become a rule (“I always freeze”: always, without exception?). Distortions: a subjective link is presented as fact (“He ignored me, so he thinks I'm useless”: how does one prove the other?). The practitioner hears the pattern and asks the question that fills the gap.
- Omission: “I'm not ready.” Ready for what, specifically? What would ready look like?
- Generalisation: “Nobody supports me.” Nobody at all? Who came closest?
- Distortion: “She's late, so she doesn't respect me.” What else could lateness mean?
- Modal operator: “I can't say no to him.” What would happen if you did?
The Milton model, rapport and calibration
The Milton model is the meta model in reverse: deliberately vague, permissive language, modelled on the hypnotist Milton Erickson, that lets a client fill in their own meaning. “You may find yourself relaxing at whatever pace suits you” gives an instruction and an escape at once. Our fourth module pairs it with the meta model and with “sleight of mouth”, a set of conversational reframes; the hypnotic use of the same language is in our guide to hypnosis inductions. It should never be used to steer a client somewhere they have not agreed to go.
Calibration and rapport underlie every other technique. Calibration, in our manual, is the close observation of the outward signs that reflect a person's internal state: breathing, skin tone, posture, voice, micro-movements, and the mismatches between them. Rapport is built by matching some of what you observe before you lead. Social psychology supports the finding that behavioural mirroring increases liking and cooperation. It does not support NLP's claim that eye movements reveal whether someone is remembering or imagining; that idea has been tested and failed, and we teach it as history.
What does the evidence say about NLP techniques?
That NLP as a system has not been shown to do what its early literature claimed, and that several of its techniques work because they are relabelled versions of things that do. A 2012 systematic review in a British general-practice journal found too little good-quality evidence to support NLP for any health outcome; a 2010 review of 33 studies reached a similar conclusion about the model's theoretical claims, eye-accessing cues included. On the other side, conditioning, cognitive reframing, structured questioning, imagery work and behavioural matching each have evidence of their own, and a practitioner who uses them for coaching goals is on reasonable ground.
What the evidence shows
Small trials report gains in confidence, communication and self-rated wellbeing after NLP-based coaching, usually with no active comparison. No condition has been shown to respond to NLP in a well-controlled trial. The techniques that overlap with cognitive-behavioural and hypnosis research carry that research's support, not their own. Practitioners should describe NLP as a communication and coaching method and let the specific tools stand on the evidence behind them.
Where does NLP end and treatment begin?
At the moment a client's difficulty has a clinical name. Anchoring a calm state before a presentation is coaching. Working on a diagnosed anxiety disorder with the same technique is practising psychology without a licence in most states. Our practitioners use these six tools for goals, habits, decisions and conversations, and ask every client at intake whether they are under medical or psychological care. The 56 teaching hours of our in-person NLP program (8 classroom days, $3,000) include a module on designing a clean coaching contract precisely so the line is drawn before the first session, not discovered in the third. How NLP sits beside hypnosis and EFT for a mind-body practice is compared in a separate article.
Scope of practice
Say “coaching”, “communication”, “goal”. Do not say “treat”, “heal”, “phobia” or “trauma” in your marketing or your notes. If a client presents a condition, or a memory that produces distress rather than reflection, the correct NLP technique is a referral. State rules on coaching are loose; state rules on the practice of psychology are not, and the difference is decided by what you claim and work on, not by what you call yourself.
Questions on this topic.
What is the most effective NLP technique?+
In practice, the meta model, because it is the one you use in every session and the one with the least dependence on belief: precise questions turn vague complaints into workable goals. Anchoring is the most reliable state tool when set at peak intensity and tested. Neither has strong NLP-specific research; both borrow their support from structured questioning and classical conditioning. The technique that works is the one matched to the problem.
Can you use NLP techniques on yourself?+
Yes, and practitioners are expected to. Self-anchoring, setting a gesture to a resourceful state you have built on your own, is taught in our first-level manual and works the same way as anchoring a client. The swish can be run alone once you know your cue image. The meta model is harder to apply to yourself, because the point is to catch language you do not notice; writing the statement down helps. Calibration needs another person.
How long does an NLP anchor last?+
Days to months, depending on how cleanly it was set and how often it is used. An anchor set at peak intensity on a pure state and fired regularly strengthens; one set late, on a mixed state, or left unused for weeks fades, as conditioned responses do. Practitioners tell clients to fire a new anchor daily for the first two weeks. An anchor that never worked was usually set after the peak.
Is NLP the same as hypnosis?+
No, though they overlap. NLP was built partly by modelling the hypnotist Milton Erickson, and the Milton model, its permissive language pattern, is used in hypnosis. Hypnosis works with a formal or conversational trance and suggestion; NLP is a collection of communication and change techniques that mostly run in ordinary conversation. Our programs are separate: 8 classroom days for NLP, 6 for hypnosis, with different scopes and different state rules on titles.
Do NLP eye-movement cues really work?+
No. The claim that looking up-left means remembering an image and up-right means constructing one has been tested in controlled studies and did not hold; people's eye movements do not reliably map to representational systems, and observers using the model do not detect lies or memory type better than chance. We teach eye-accessing cues as history and warn students not to use them. Calibration of breathing, posture and voice, by contrast, is careful observation and worth practising.
What is the difference between the meta model and the Milton model?+
They are mirror images. The meta model recovers information: it asks precise questions to fill in what a vague statement omits, over-generalises or distorts. The Milton model deliberately removes information: it uses vague, permissive language so the listener fills in their own meaning, which is useful for relaxation, suggestion and gentle reframing. Practitioners use the meta model to understand a client and the Milton model to guide attention. Neither is a treatment.
Can NLP techniques be harmful?+
The techniques themselves are low-risk; the misuse is not. Two dangers. Applying reframing or anchoring to a traumatic memory can bring it forward without the safety frame a licensed professional would provide. And the Milton model, used to steer a client toward a decision they did not choose, is manipulation, whatever it is called. Our program teaches consent, a written coaching contract and referral rules as part of the technique, not as an afterthought.
How many NLP techniques are there?+
Depends on who is counting. The standard practitioner curriculum names 40 to 60 patterns, and the big reference books list more, but almost all are combinations of six operations: anchoring states, reframing meaning, shifting submodalities, questioning with the meta model, suggesting with the Milton model, and calibrating and matching. A practitioner who can do those six fluently can reconstruct most of the rest. A certificate that promises 100 techniques in seven days is promising vocabulary, not skill.
Tags:
NLPWhat it is
About the author
Valérie Fabre · Founder & Director, Harmonika Institute
Valérie Fabre founded Harmonika Institute in 2025, after a decade seeing clients in private kinesiology practice and, before that, thirteen years in international trade. A French-born polyglot who now splits her time between Dallas and Miami, she sets the school's curriculum and editorial standards and champions ethical, in-person training.
Read more about Valérie Fabre →