AI Guides › Prompting

The Difference Between A Constraint And A Preference In Your Prompt

Name: JulieMango AI Agent Marketplace Plans
Brand: JulieMango

By Nigel Guy · 2 min read

People write prompts as a flat list of things they'd like, and every item on the list gets treated with the same weight — which means the one thing that actually can't move (the word limit, the required disclaimer, the format the recipient expects) gets no more emphasis than the thing you'd genuinely be fine trading away (whether it opens with a question or a statement). When everything is stated the same way, everything gets treated as negotiable.

The rule: say plainly which parts of the request are fixed and which are just leanings — a constraint that reads like a preference will get quietly traded away the moment something else in the prompt pulls against it.

The distinction

	Constraint	Preference
Definition	Must hold, no exceptions	A default you'd like, but would trade off
How to state it	"Must be under 200 words" / "never include pricing"	"I'd lean toward a warmer tone, but use judgement"
What happens if ignored	The output is unusable for its purpose	The output is fine, just not your first pick
How many you should have	As few as truly matter — three or four	As many as you like; they're low-cost

The mechanism

Sort your requirements into the two columns above before you write the prompt, not while writing it. Most people discover a "constraint" was really a preference only when they're annoyed the model didn't honour it.
Say the word. Literally write "must" or "non-negotiable" for constraints, and "prefer" or "lean toward" for preferences. The label does real work — it's the difference between a rule and a suggestion.
Put constraints first, separately, before the descriptive part of the request. A constraint buried in the middle of three paragraphs of context reads like ambient detail, not a rule.
Limit constraints to the ones that would actually make the output unusable if broken. Everything else is a preference, however strongly you feel about it in the moment.

What to skip

Skip labelling every detail a constraint out of habit — a prompt where everything is "must" is functionally the same as a prompt with no constraints at all, because nothing stands out. And skip re-stating preferences as constraints just to get more of what you want; if it isn't actually a dealbreaker, say so, and accept the output might land somewhere else within that range.

Guardrails

Stating a constraint clearly improves the odds it's honoured; it isn't a guarantee. Check the output against your actual must-haves before using it.
A prompt with too many "musts" is a sign the request needs splitting into two smaller ones, not a sign you need to phrase the constraints harder.
This is about how you write the request, not about what the model can or can't technically enforce — behaviour there varies and is worth checking against the current guidance for whatever tool you're using.

All 751 AI guides · JulieMango plans from £17/mo