The GOAL Framework: How to Brief AI Properly in Fifteen Seconds

In my last post I made a case that I still believe: you get weak answers from AI because you brief it in six words, like a search box, when you should brief it like you’d brief a person. Give it a role, the context it’s missing, a precise ask, and a standard for what “good” looks like.

And almost everyone who nods along hits the same wall about an hour later.

I don’t have time for that. It’s 11pm, the deck is due, and writing a little four-part essay before every single prompt is a tax I’m not going to pay forty times a day. So they drift back to the one-liner, get the flat answer, and quietly conclude the structure thing is nice in theory.

They’re not wrong to push back. If good prompting requires a ceremony, nobody will do it under deadline. So the real question isn’t should you give AI structure — it’s how do you make structure so fast it costs you nothing?

That’s what GOAL is for.

GOAL is the fifteen-second version

GOAL is the framework I reach for more than any other. Not because it’s the most powerful — it isn’t — but because it’s the one I can run without thinking, dozens of times a day. It’s the daily driver: quick analyses, standard deliverables, the ordinary work that fills most of your hours.

It’s four questions. You answer them in your head, or in four short lines, before you hit enter.

G — Goal. What do you actually want to achieve? O — Output. What format do you need it in? A — Audience & Context. Who is this for? L — Leverage. What constraints or style should shape it?

That’s the whole thing. No memorising, no template to open. Four questions you can ask yourself faster than you can type the lazy version. And each one closes off a specific way the answer usually goes wrong.

The same request, run through GOAL

Let me use the exact example I flagged last time as a wish rather than a brief: “Give me a market entry strategy for Southeast Asia.”

You already know what you get back — a competent, generic essay about a region the size of a continent, hedged in every direction because the model has no idea who’s asking or why. It isn’t wrong. It’s just useless to you.

Now watch what four lines do to it.

Goal: Create a market entry assessment for Southeast Asia. Output: Executive briefing document with clear recommendations. Audience: A private equity investment committee — financially sophisticated. Leverage: Focus on ROI, two-page maximum, include the risks.

Same request. Same thirty seconds of typing, near enough. But look at how much you’ve told it. It now knows this isn’t a student essay — it’s a decision document for people who read balance sheets for a living. It knows the answer has to fit on two pages, which forces prioritisation instead of a data dump. It knows to lead with returns and to be honest about what could go wrong, because a PE committee that doesn’t see the risks flagged will assume you missed them.

The output stops being an encyclopedia entry and starts being something you could actually put in front of the committee. You’ll still sharpen it — that’s your job — but you’re editing a draft, not commissioning one.

The two letters everyone skips

If GOAL only ever nudged you to state your goal, it wouldn’t be worth naming. Most people are half-decent at saying what they want. The value is in the two letters they leave out: Output and Audience.

Skip Output and the model defaults to prose — paragraphs, when you needed a table, or a one-line answer, or a five-slide structure. Naming the format is the single cheapest way to save yourself a reformatting job later. “Give me a comparison table,” “draft this as an email,” “structure it as three slides” — the shape changes before a word is written.

Skip Audience and the model writes for nobody in particular, which means it writes for everybody, which means it writes for the blandest common denominator. The same analysis pitched to a PE committee, a nervous first-time founder, and your own managing partner should come out in three genuinely different registers. The model can do all three effortlessly — but only if you tell it which room it’s standing in.

Those two questions take four seconds to answer and they move the output more than anything else in the prompt. That’s the whole bet of GOAL: the cheapest inputs do the heaviest lifting.

When GOAL is enough — and when it isn’t

GOAL is a workhorse, not a Swiss Army knife, and part of using it well is knowing its edge.

Reach for GOAL when you’re moving fast and the stakes are ordinary: a first-pass analysis, a summary, an email, a bit of structured thinking to get unstuck. That’s the great majority of what any of us does in a day, and for all of it, GOAL is plenty. Speed with quality is exactly the trade it’s built for.

Step up when the work has to leave the building and survive scrutiny. When you’re drafting a genuine client deliverable — something that has to be rigorous, defensible, and nuanced — that’s when I move to a heavier framework built for professional depth. (In the last post that was RCAS: Role, Context, Ask, Style. Same idea as GOAL, more room for craft.) And when the task is real strategy, or a hard message that has to persuade, or a piece of research that could sprawl for days, there are purpose-built shapes for each of those too.

The skill isn’t loyalty to one framework. It’s recognising, in the first two seconds, which kind of job you’re actually holding — and reaching for the right shape without breaking stride.

Make it a reflex

Here’s the honest truth about why this works: it isn’t clever, it’s just consistent. The gain doesn’t come from any single brilliant prompt. It comes from never again typing the lazy version.

So don’t try to memorise anything. For your first week, keep the four words somewhere you can see them — a sticky note, the top of a scratch file, wherever your eyes land when you open a chat. Goal. Output. Audience. Leverage. Glance at them, answer them, hit enter. After a week or two you’ll stop glancing, because the questions will have moved into the part of your brain that briefs people without thinking about it. That’s the whole goal — to make good prompting the path of least resistance instead of a discipline you have to summon.

Fifteen seconds, four questions, on every prompt that matters. It is genuinely the highest-return habit I know of in this whole space.

Where to go next

If you want the rest of the system, I’ve made it free. GOAL is the daily driver, but it’s one of six frameworks — one each for professional deliverables, strategy, communication, research, and innovation — and you can get all six here. The hundred ready-to-run prompts built on them, each tagged with its framework and template, come with the book as the companion Prompt Vault for readers.

And if you haven’t read it, the piece this one builds on — why consultants should prompt with frameworks instead of one-liners — is the argument underneath all of this.

But you don’t need any of that to start. You need four words and the next prompt you were about to type. Answer Goal, Output, Audience, Leverage first. You’ll feel the difference on the very first try.