The NLP Meta Model
People rarely say what they mean. They say a compressed, edited version of it — and then live inside the compression. The Meta Model is a set of questions that recovers what the sentence left out, and hands the speaker back the detail they deleted from their own thinking.
In Short
The Meta Model is a questioning framework in Neuro-Linguistic Programming that recovers the information a person has deleted, generalised or distorted when they speak. It sorts everyday language into recognisable patterns — “nobody listens to me”, “I can’t say anything”, “she thinks I’m not up to it” — and gives you one precise question for each. Ask it, and the speaker recovers the missing detail themselves. The purpose is not to prove anyone wrong. It is to help them think clearly enough to act.
Written And Reviewed By
Founder And Lead Trainer, NLP Limited
Rajiv Sharma is an NLP Master Trainer certified through The Society of NLP (USA), trained and coached directly by Dr. Richard Bandler, co-creator of Neuro-Linguistic Programming. He holds the ICF PCC credential and created the MARK Model® and the LeadFORTH framework. He is the author of Make a MARK in Life and AI-ENABLE for Sales. Over a career beginning in 1990, he has trained more than 850,000 professionals across 57 countries. His work is endorsed by Marshall Goldsmith, Brian Tracy and John Mattone.
Why It Works
Everything a person knows about a situation sits inside them in full — the sequence of events, the tone of voice, the moment it turned, the part that went well. NLP calls that the deep structure. What comes out of their mouth is a compressed extract of it: the surface structure.
Surface Structure — What Was Said
“The client was difficult.”
Deep Structure — What Was Actually There
Nothing dishonest happened here. Language cannot carry an entire experience — and nobody has the time to let it. Compression is what makes speech possible. The trouble begins when the speaker loses sight of the compression, stops treating the summary as a summary, and starts acting on it as though it were the situation. “The client was difficult” produces a plan. “The deadline was unrealistic and I felt underprepared” produces a very different one.
The Meta Model works because the missing information has not gone anywhere. It is still in the speaker. The question does not supply an answer — it points at the hole in the sentence and lets the person look into it themselves. Most of the useful work in a coaching conversation happens in the silence after a well-placed Meta Model question, when someone properly hears their own sentence for the first time.
Where it came from. The Meta Model is the first technique in NLP — it appears in the founding text, The Structure of Magic, built by modelling how two exceptional therapists actually questioned their clients. Everything else in the toolkit was constructed afterwards. If you learn one NLP skill properly, this is the one.
New to the subject? Start with What Is NLP?
The Three Transformations
Three things happen between an experience and the sentence describing it. Every Meta Model pattern is a species of one of these three. Learn the three, and the twelve arrange themselves.
Information that was there is left out of the sentence.
They said:
“They rejected my idea.”
What went missing:
Ask:
“Who specifically rejected which part of your idea?”
An interpretation is added, then treated as though it were the event.
They said:
“She didn’t reply, so she doesn’t value me.”
What is actually known:
Ask:
“How does a missing reply mean she doesn’t value you? What else could it mean?”
One or two experiences become a permanent rule.
They said:
“I always fail in interviews.”
What the rule ignores:
Ask:
“Always? Has there been a single one that went reasonably well?”
Generalisation is not the enemy. It is how anyone learns anything — touch one hot stove, avoid all hot stoves. The Meta Model challenges a generalisation only when it has started to cost the person choices. A rule that protects you stays. A rule that imprisons you gets questioned.
The Framework
Three families. Twelve patterns. One question each. Learn to hear the pattern and the question arrives on its own.
| Pattern | Sounds Like | Challenge Question |
|---|---|---|
| Simple Deletion Information dropped entirely. | “I’m upset.” | About what, specifically? |
| Comparative Deletion A comparison with nothing to compare against. | “This is the worse option.” | Worse than what? |
| Unspecified Noun Who or what is never named. | “They don’t support us.” | Who, specifically? |
| Unspecified Verb The action has no shape. | “He mishandled it.” | Mishandled it how, specifically? |
| Pattern | Sounds Like | Challenge Question |
|---|---|---|
| Universal Quantifier Always, never, everyone, nobody, every time. | “Nobody listens to me here.” | Nobody? Has there ever been one single time someone did? |
| Modal Operator Of Necessity Must, should, have to, ought to. | “I have to say yes to this.” | What would happen if you did not? |
| Modal Operator Of Possibility Can’t, impossible, unable. | “I can’t raise it with her.” | What stops you? |
| Lost Performative A judgement with no judge attached. | “It’s unprofessional to push back.” | According to whom? Who says so? |
| Pattern | Sounds Like | Challenge Question |
|---|---|---|
| Nominalisation A process frozen into a noun. | “There’s no communication in this team.” | Who is not communicating what, to whom? |
| Mind Reading Claiming to know another’s inner state. | “She thinks I’m not up to it.” | How do you know that? |
| Cause And Effect Another person’s act made you feel. | “His tone makes me anxious.” | What does he do, and how do you respond internally? |
| Complex Equivalence X means Y, with no proof of the link. | “He didn’t reply, so he’s written us off.” | How does not replying mean he has written you off? |
Why some books say nine, and others say thirteen. Different NLP schools split and merge these differently — presuppositions are sometimes counted separately, simple and comparative deletion are sometimes folded together, and unspecified nouns are sometimes called referential indices. The count varies. The listening does not. Learn to hear what is missing and the number becomes an administrative detail.
Interactive Tool
Write a sentence you keep hearing — from a colleague, a client, a team member, or from yourself. The Workbench finds the patterns hiding inside it and gives you the exact questions to ask next.
Step 1 Of 3
Write it exactly as you heard it. The precise words are the whole point.
Or start from an example:
Write or choose a sentence first.
Step 2 Of 3
Here is what we can hear in it. Untick anything that does not fit — you know the situation, we do not.
Nothing obvious surfaced — which usually means one of two things. Either the sentence is already specific (good), or the vagueness sits in a word only you can hear. Write the version of the sentence that actually bothers you, rather than the tidy one.
Your Result
One question at a time. Ask it, then let the silence work. The value sits in what they say next, not in what you ask.
A Meta Model question asked cold reads as an interrogation. Put a frame in front of it and the same question lands as care: “I want to make sure I’ve understood — when you say ‘nobody’, who do you mean?” Rapport comes first, always. See how to build rapport.
Knowing the questions is one thing. Asking them live, under pressure, without damaging the relationship is a skill built with feedback. Send your result to Coach Rajiv Sharma and he will point you the right way.
This is what will be sent:
Your result is private to you — nothing was stored on this page. Sharing it is entirely your choice.
Step By Step
The Meta Model is a challenge, however gently phrased. Without rapport it reads as attack, and the person defends the sentence instead of examining it. Match their pace, show you are on their side, and earn the right to ask.
Most people listen for content and respond with advice. Train yourself to listen one level above the content — for the always, the can’t, the she thinks. The pattern tells you where the missing information sits.
A single sentence often carries three or four patterns. Asking about all of them turns a conversation into a cross-examination. Pick the one carrying the most weight — usually the belief the person is most attached to.
“Help me understand…”, “I want to be sure I’ve got this right…”, “Can I ask you something specific?” The frame carries the intent; the question carries the precision.
The silence after the question is where the work happens. Let it sit. The moment you fill it, you have taken the search off them and put it back on yourself.
Their answer will contain new patterns. Follow those. The Meta Model is a way of tracking vagueness wherever it moves, rather than a script to be run through.
You are chasing the point at which a person can do something, not perfect precision. When the sentence becomes a specific situation with a specific person and a specific next step, stop.
Phrasing
The technical name for a pattern is for you. It is never for the person you are speaking to. Nobody has ever been helped by being told they produced a complex equivalence. Here is the same question in language a human being can receive.
“What is your evidence for that complex equivalence?”
“What makes those two things mean the same to you?”
“What is the missing referential index?”
“Who specifically are you thinking of?”
“What is the deleted comparison there?”
“Better compared with what?”
“That is a universal quantifier. It cannot be true.”
“Has there been even one time it went differently?”
“You are nominalising the process there.”
“Who would be doing what differently, if that changed?”
The pattern name is your map. The question is what you say out loud. Keeping the two apart is most of the difference between a practitioner people open up to and one they learn to avoid.
The Line You Must Not Cross
A Meta Model question can rescue a conversation or end one. The words barely change. What changes is whether the person feels examined or feels understood. Move the switch.
They Said
“Nobody understands me.”
“Nobody? That cannot be true. Who exactly?”
The pattern was spotted correctly. The person now feels corrected, caught out, and slightly stupid. They will defend the sentence rather than examine it — and they will be more careful about what they tell you next time. You won the point and lost the conversation.
A Meta Monster is the informal name for someone freshly trained who challenges every vague statement they hear. Someone says “I had a terrible day” and gets back “Terrible compared with what? Which part? How do you know?”
Technically, every one of those questions is valid. Socially, the person has become exhausting, and their family has quietly stopped telling them things. Skill in NLP has nothing to do with how many patterns you can name. It is choosing the right question, at the right moment, with the right intention — and knowing when the right number of questions is zero.
Where It Earns Its Place
A Result From The Field
Diners Club trained every manager and contact centre agent in NLP communication skills — precision questioning at the centre of it. Agents stopped responding to the surface complaint and started recovering what the customer actually meant. Net production rose 254% in customer spending and customer loss fell 67%. The contact centre, until then a cost centre, was reclassified as a revenue-producing unit.
“It’s too expensive” is a comparative deletion, and almost never about price. Too expensive compared with what? recovers the benchmark the buyer is actually using — and that benchmark is the real conversation. Applied across a sales floor, this is what NLP sales training is built on.
“The team’s morale is low” is a nominalisation wrapped around an unspecified noun. Who, specifically, is feeling what? turns an unsolvable mood into three named conversations you can have this week. Core to leadership development.
“Your performance needs to improve” tells an employee nothing they can act on. The Meta Model works in both directions here — the manager should ask it of their own feedback before delivering it. Which part, and what would good look like?
Almost every limiting belief is a modal operator or a complex equivalence. What stops you? and How does that mean that? do more coaching work than any amount of encouragement. Trained formally in ICF coach certification.
Mind reading drives most workplace conflict: two people each certain they know what the other intended. How do you know that? is often the only question the situation needs.
The hardest and most useful application. Write down the sentence you have been telling yourself, then challenge it as though a colleague had said it. Most of it will not survive.
Interactive Tool
Reading the patterns takes an hour. Hearing them in live speech takes a week of deliberate listening. One pattern a day — listen only for that one, ask nothing, write down what you catch. Tick each day as you complete it.
0 of 7 days complete
Tick a day as you complete it. When you have finished, send your log to Coach Rajiv for feedback on what you noticed.
The Twin
They are opposites, and deliberately so. One sharpens language until the situation becomes workable. The other loosens it until the listener supplies their own meaning. Practitioners learn both, and choose.
| The Meta Model | The Milton Model | |
|---|---|---|
| What it does to language | Makes it specific | Leaves it open |
| Direction | Chunks down, towards detail | Chunks up, towards possibility |
| Sounds like | “What kind of pressure, specifically?” | “You may begin to notice new ways of responding.” |
| Who supplies the meaning | The speaker recovers their own | The listener creates their own |
| Used for | Clarity, problem-solving, discovery, feedback | Reflection, imagination, opening possibility |
| Failure mode | Interrogation | Waffle |
| Remember it as | Be more specific | Leave space for interpretation |
Same statement, two treatments. “I can’t handle pressure.” The Meta Model asks: “What kind of pressure are you referring to?” The Milton Model offers: “And you might find there are kinds of pressure you already handle without thinking about it.” Neither is superior. They do different jobs. Our full guide to the Milton Model is in preparation.
Refining The Skill
Cause: Questions asked without a frame, back to back, with no rapport underneath. Fix: One question per exchange. Frame it. Then listen for twice as long as you spoke.
Cause: The person hears the question as a challenge to them, rather than to the sentence. Fix: Attach the curiosity to yourself — “I might be missing something here” — so the gap in understanding is yours, not a fault of theirs.
Cause: Chasing specificity for its own sake. The Meta Model can recover detail forever. Fix: Keep the outcome in view. Ask what the person needs to be able to do, and stop when they can do it.
Cause: The vagueness is protecting something. A person who stays vague about a problem usually has a reason. Fix: Move to the positive intention behind the vagueness rather than pressing harder on the sentence.
Cause: Precision applied where presence was needed. Fix: A distressed person needs to be met before they are questioned. Listen, acknowledge, help them feel understood. State first, structure second. See state management.
Cause: The classic post-training phase — challenging every sentence at the dinner table. Fix: The skill is knowing when not to ask. Most conversations want warmth, not precision. Save it for where it changes something.
Twenty-One Questions Answered
Every question, with the answer in full. Filter by what you came for — or read all twenty-one in order.
The NLP Meta Model is a set of questions used to make unclear language specific. People speak in shortcuts — “nobody respects me”, “I cannot succeed”, “my manager makes me nervous”. These statements feel true to the speaker, and they leave out almost everything that matters. The Meta Model supplies the question that recovers what was left out: who specifically does not respect you?, what stops you?, how exactly does he make you nervous? The purpose is never to prove someone wrong. It is to help them think clearly enough to act.
To recover missing information, question assumptions, and give the person back their choices. Take “I can’t speak confidently in meetings.” Ask “what specifically stops you?” and the person may discover it is a fear of being judged, or inadequate preparation, or one particular senior leader in the room, or a single bad experience years ago. Four completely different problems, each with a different solution — and all four were hidden inside one sentence. The Meta Model turns a vague problem into a workable one.
By listening to the words rather than the story. When someone says “the presentation went badly”, the sentence conceals which part went badly, according to whom, what actually happened, what was expected instead, and whether anything went well. One question — “which part did not go as you expected?” — moves the person from a broad judgement towards something they observed. Judgements cannot be worked on. Observations can.
Deep structure is the complete experience as it exists inside the person. Surface structure is the shortened version that comes out in words. Someone says “the client was difficult” — that is surface structure. Underneath it: the client changed the requirements, spoke impatiently, the deadline was unrealistic, and the person felt underprepared and has told nobody. The Meta Model moves the conversation from the short statement towards the fuller experience. This is not an accusation of dishonesty. Language simply cannot carry an entire experience, and it never has.
The three ways people compress an experience into a sentence. Deletion leaves information out: “they rejected my idea” omits who, which idea, and what they actually said. Distortion adds an interpretation and treats it as a fact: “she didn’t reply, so she doesn’t value me” — the missing reply is known, the reason is invented. Generalisation turns a few experiences into a rule: “I always fail in interviews.” Every Meta Model pattern belongs to one of these three.
They fall into three families. Deletions: simple deletion, comparative deletion, unspecified noun, unspecified verb. Generalisations: universal quantifiers, modal operators of necessity, modal operators of possibility, lost performatives. Distortions: nominalisation, mind reading, cause and effect, complex equivalence. Each has one matching challenge question, listed in full in the table above.
It depends on who taught you, and the disagreement is real. Most schools teach twelve. Some teach nine by merging the deletions. Some teach thirteen by counting presuppositions separately. The original work lists them differently again. None of this matters in a live conversation. What matters is whether you can hear that something is missing and ask for it. Count the patterns however your training counts them, and then spend your attention on the listening instead.
The seven that will carry you through most conversations:
The best question is the one that helps the person become more specific without feeling attacked.
A nominalisation is a living process frozen into a noun — communication, leadership, respect, trust, performance, morale. The test: if you cannot put it in a wheelbarrow, it is probably a nominalisation. It matters because a frozen noun feels permanent and unfixable, while the process underneath it can always be changed. “There is no communication in our team” is hopeless. “Who is not telling whom what?” is a meeting you can hold on Thursday. Turning the noun back into a verb is the whole intervention.
The Meta Model makes language specific. The Milton Model uses language broadly and indirectly. Given “I can’t handle pressure”, the Meta Model asks “what kind of pressure, specifically?” The Milton Model offers “you may begin to discover new ways of responding.” One clarifies; the other leaves room for the listener to build their own meaning. Remember it as: Meta Model — be more specific. Milton Model — leave space for interpretation. Both are useful. They do different jobs.
They share a purpose — examining an assertion rather than accepting it — and differ in method. Socratic questioning follows the logic of an argument. The Meta Model works from the surface structure of the sentence itself, using grammar as the map to what is missing. It is more systematic, and considerably faster to learn.
Richard Bandler and John Grinder, as NLP's founding technique, set out in The Structure of Magic. It was built by closely modelling how two exceptional therapists questioned their clients, and reverse-engineering the language patterns behind their results. Every other NLP technique was constructed afterwards.
Yes — when used carelessly. Too many questions, asked coldly, leave a person feeling interrogated, judged, corrected and unsafe. Someone says “nobody understands me.” The harsh response: “Nobody? That can't be true. Who exactly?” The better response: “That sounds hard. Who do you most wish understood you?” Both seek clarity. Only one keeps the person in the room. The Meta Model should feel like exploration, never cross-examination.
Listen before questioning. Acknowledge the feeling first. Keep a calm, curious tone. Ask one question at a time. Leave most sentences unchallenged. Stop as soon as enough clarity has emerged. And never use technical NLP terminology with the person — nobody has ever been helped by being told they produced a complex equivalence. Asked with real curiosity, a Meta Model question feels like being taken seriously. Asked to win a point, it feels like an attack.
When the meaning is unclear. When someone is stuck inside a broad conclusion. When an assumption is being treated as a fact. When important information is missing. When the person's own language is shrinking their choices. And when clearer information would lead to a better decision. Not when someone is highly emotional — listen, acknowledge and help them feel understood first.
Stop when the person has identified the real issue. When there is enough information to act. When the conversation starts going in circles. When they begin to feel defensive. When the next question would add nothing. And when emotional support has become more valuable than further analysis. The goal was never to collect every detail — it was to create enough clarity for a better choice.
An informal name for someone who challenges every vague statement they hear. “I had a terrible day.” “Terrible compared with what? Which part? How do you know?” Technically the questions are valid. Socially the person has become unbearable, and their friends have quietly stopped telling them things. Good NLP has nothing to do with displaying how many patterns you can identify. It is choosing the right question, at the right time, with the right intention.
They reduce misunderstanding at the point where it starts. A manager says “your performance needs to improve” and the employee has no idea what to change. One question fixes it: “which part of my performance, and what would good look like?” Used well, the Meta Model helps people understand what someone really means, separate fact from assumption, clarify expectations, reduce needless conflict and make better decisions. It earns its keep whenever words like better, professional, supportive, difficult or successful are used with no explanation attached.
In coaching, a client says “I don’t have the confidence to change careers.” The coach asks “what specifically would you need in order to feel prepared?” — and a fixed problem becomes a list of practical steps. In leadership, an employee says “the team is not cooperating.” The leader asks “what are people doing, or not doing, that you describe as a lack of cooperation?” — and now the leader can address behaviour rather than label people. In both cases the aim is identical: help the person think, rather than think for them.
Start by noticing vague words in ordinary conversation — always, never, everyone, nobody, can’t, must, should, better, worse, difficult, successful, rejected. Then hold the questions in your head without speaking them: Who specifically? What specifically? How specifically? According to whom? Compared with what? What stops you? How do you know? What would happen if you did? The seven-day drill on this page gives you one pattern per day. Take the statement “my manager doesn’t support me” and explore it: what kind of support? what has your manager done, or not done? when do you feel unsupported? has your manager ever supported you? have you clearly asked for what you need? Choose the one question that creates the most clarity, rather than asking all five.
The patterns, yes — most people can recognise them within an hour of reading. What cannot be self-taught is the live skill: asking with rapport, under pressure, choosing which of four patterns to challenge in real time, and hearing your own tone as the other person hears it. That last one is the reason. Nobody can hear their own questioning accurately, which is why practitioners train with someone listening.
Where To Go Next
Reading the twelve patterns takes an hour. Hearing them in real time, choosing the right one, and asking it so the other person opens up rather than closes down — that is built with practice and feedback. Train the Meta Model with an NLP Master Trainer on a certified NLP programme, or bring it to your team through corporate training.