Chunking up and down means deliberately changing the size of the idea you are talking about. Chunking up moves to the larger category or the purpose behind a thing (“what is this an example of?”), chunking down moves to the concrete detail underneath it (“what specifically?”), and chunking laterally moves sideways to another example of the same size. In Neuro-Linguistic Programming these three moves are used on purpose: you chunk up to find agreement, down to build a plan, and sideways to find an alternative when the first one is blocked.
| Move | Question that triggers it | What you get |
|---|---|---|
| Chunk up | What is that an example of? What is this for? | Shared ground, meaning, agreement |
| Chunk down | What specifically? Which one? How exactly? | Detail, evidence, a first action |
| Chunk laterally | What else is an example of the same thing? | Alternatives, metaphors, reframes |
What does chunking up look like in a real sentence?
Chunking up replaces a specific item with the class it belongs to, or with the outcome it serves. The classic ladder runs like this: a Volkswagen Golf is a car, a car is a vehicle, a vehicle is transport, transport is freedom of movement. Every step up loses detail and gains agreement. Two people can argue for an hour about the Golf and still nod at “we both want to get around easily”.
That is why chunking up is the move for stuck disagreements. Watch it in a short exchange:
- “I want to work from home three days a week.” / “I need the team in the office.” Both positions are small chunks and they collide.
- Chunk up: “What would working from home give you?” Answer: focused time. “And what does having everyone in the room give you?” Answer: people asking each other questions instead of waiting a day.
- One level higher, both are asking for the same thing: work that moves without friction. Now there is something to design around, which was invisible one chunk lower.
Two questions do almost all of the upward work. “What is that an example of?” moves you up the category ladder. “What will that do for you?” moves you up the purpose ladder. The second one is the one used in outcome work and goal setting, where a stated goal is tested against the intention behind it before anyone spends effort on it.
What does chunking down look like?
Chunking down takes a large, comfortable word and asks it to become something you could film. It is the direction most conversations avoid, because big words feel like agreement and detail is where the work starts.
| Big chunk someone says | Question | Small chunk that comes back |
|---|---|---|
| “We need better communication.” | Better how, between whom, in which situation? | “Handover notes on Friday so Monday does not start with guessing.” |
| “I want more confidence.” | Confidence doing what, in front of whom? | “Speaking first in the Tuesday meeting instead of third.” |
| “They never support me.” | Who specifically, and what would support look like? | “My manager, saying in the meeting that the delay was not mine.” |
| “The launch went badly.” | Which part of it, measured how? | “Sign-ups were fine, onboarding emails went out two days late.” |
If those questions look familiar, they should. Chunking down and the Meta Model overlap almost completely: the Meta Model is the precise set of questions that recovers what a sentence left out, and every one of them makes the chunk smaller. If you want the full question set rather than the principle, the article on Meta Model questions with examples lists them one by one with the phrasing that does not sound like an interrogation.
How do you chunk sideways?
Lateral chunking goes up one level, then comes back down a different branch. A car chunks up to vehicle, and vehicle chunks back down to bicycle, train, boat. You never say the middle step out loud. What the other person hears is a fresh option or a metaphor that arrived from nowhere.
Examples of the sideways move in ordinary use:
- “We cannot afford the conference.” Up to “ways to meet the people we would have met there”, back down to a small dinner, a podcast appearance, three well-chosen emails.
- A client insists the only route is a promotion. Up to “work that uses what I am good at”, back down to a lateral move, a project lead role, a change of team.
- Every useful metaphor is a lateral chunk. “Your calendar is a wardrobe, not a warehouse” works because it went up to “containers with a fixed size” and came back down somewhere else.
Which direction should you use, and when?
A simple rule covers most cases. If people are arguing, chunk up until they agree. If people are agreeing on something vague, chunk down until they are arguing about something real. If people are stuck with one option, chunk sideways.
Two signals tell you the chunk size is wrong. When you hear a lot of nouns that have no picture attached, growth, alignment, transparency, culture, you are too high and nothing will happen until you go down. When you hear a long list of details and nobody can say why any of it matters, you are too low and the conversation needs one step up before it can be prioritised. Mismatched chunk size also explains a good share of failed rapport: a person who thinks in outcomes talking to a person who thinks in steps will each find the other frustrating, and both will call it a personality clash.
A ten-minute chunking drill you can run today
Take one sentence you have said recently that has not led anywhere. Write it at the middle of a sheet of paper, then work through these five steps.
- Chunk up twice. Ask “what will that do for me?” and write the answer above the sentence. Ask it again of the answer. You now have the purpose the sentence was serving.
- Chunk down three times. Ask “what specifically?” of the original sentence, then of each answer, until you reach something with a time, a person and a verb in it. That is the level where action lives.
- Chunk sideways once. From the purpose at the top, find a second route down that is not the original sentence. Write it next to the sentence.
- Compare the two routes. Ask which of them costs less, and whether you were attached to the first one for reasons that have nothing to do with the outcome.
- Keep the bottom line. Copy the smallest chunk from step two into your calendar. The rest of the sheet was thinking, that line is the change.
Done on paper this takes ten minutes. Done out loud with a partner who only asks the questions and does not comment, it takes five and works better, because you hear your own vague words as you say them. That is also how it is practised in a training room: one person speaks, one person asks, and a third watches which direction actually moved things. The seven-day NLP Practitioner training in Lisbon runs that loop repeatedly in groups of no more than ten, which is the only way this stops being a diagram and becomes a habit.
Where chunking goes wrong
Three honest warnings, because the technique is easy to misuse.
Chunking up can be used to dodge. Every complaint dissolves if you take it high enough, and a manager who answers “the report was late again” with “we all want the company to succeed” is not building agreement, they are avoiding the point. Up is for finding shared ground before you come back down, not for staying there.
Chunking down can turn into cross-examination. Four “what specifically?” questions in a row feel like being audited. Space them out, put them inside normal sentences, and pay attention to whether the person is still with you. Calibration, noticing the small changes in breathing, colour and voice, is what tells you when to stop asking.
And chunking is a language pattern, not a treatment. It clarifies what someone means and where they could act. It does not resolve trauma, grief or a clinical condition, and it complements professional care rather than replacing it.
Carolin Mallmann, Licensed Trainer of The Society of NLP, Lisbon
FAQ
What is the difference between chunking up and chunking down in NLP?
Chunking up moves to a bigger category or to the purpose behind something, so detail drops away and agreement becomes easier. Chunking down moves to the specific detail underneath, so the topic becomes concrete enough to act on. Both are the same skill pointed in opposite directions, and the questions that trigger them differ: “what is this an example of?” goes up, “what specifically?” goes down.
Can you give a quick example of chunking up and down on the same word?
Take “holiday”. Chunk up and you get travel, then rest, then a life that does not need escaping from. Chunk down and you get two weeks in September, the Algarve, a flat with a kitchen, a flight booked before prices rise. The upward chain tells you why the holiday matters. The downward chain is what you can book.
Is chunking the same as the Meta Model?
They overlap but are not identical. The Meta Model is a specific set of questions for recovering information that a sentence deleted, distorted or generalised, and all of them chunk down. Chunking is the wider principle, and it includes the upward and sideways moves that the Meta Model does not cover. The Milton Model, by contrast, chunks up on purpose by staying vague.
How do I chunk up without sounding evasive?
Name the detail before you leave it. “So the handover on Friday did not happen, and I want to sort that. Before we do, can I check what we both want Monday mornings to look like?” The person hears that their point landed, which is what makes the move upward feel like progress instead of a dodge. Then come back down and settle the original item.
What is lateral chunking used for?
It generates options and metaphors. You go up one level from the blocked item, then come back down a different branch, which produces something that serves the same purpose by another route. Most good analogies in coaching and teaching are lateral chunks, built from the same purpose as the thing they explain.
How long does it take to get good at this?
Recognising chunk size in someone else’s speech takes a few weeks of deliberate listening. Choosing a direction mid-conversation, while also holding the relationship and your own point, is the part that needs supervised practice, because in real time you notice the vague noun a sentence too late. Most people get there faster with someone watching and naming what they missed.
Keep exploring
- Meta Model questions: examples you can actually use, the full downward question set, with phrasings that keep people talking.
- Logical levels exercise: find the level that is stuck, a second way of changing altitude, this time through six levels of change rather than categories.
Carolin Mallmann is a Licensed Trainer of The Society of NLP, trained by Richard Bandler. She has trained more than 300 people and runs English-language NLP trainings in Lisbon, in groups of no more than 10.
