Prompting with frameworks the model already knows

Talking to a model is like briefing someone with no memory and no idea who you are. When you brief a colleague you give them background, say what you want, and adjust your language to suit them. A model needs all the same things, just written down. That’s all structured prompting is.

Structure with MECE

I don’t usually go in for acronyms, but MECE earns its place: Mutually Exclusive, Collectively Exhaustive. Break the prompt into parts that don’t overlap and together cover everything the model needs.

1. Context. The background and the scene.
2. Goals. What you’re trying to achieve.
3. Audience. Who it’s for.
4. Style and tone. How it should read.
5. Rules. Specific dos and don’ts.
6. Additional information. Anything that doesn’t fit above.

The point is that each part tells the model something it needs and leaves out things it doesn’t.

Front-load the important context

Useful quirk: a lot of models weight what appears earlier in the prompt more heavily. If something is critical to the task, put it near the top rather than at the end. Anthropic’s prompt-engineering course is worth working through if you want to go deeper.

Models also cope well with structure, so XML-style tags help them parse a document and let you point at specific sections in follow-up messages:

<document>
  <section name="overview">This is the overview.</section>
  <section name="features">A list of features.</section>
  <section name="pricing">Pricing information.</section>
</document>

Borrow a framework the model knows

Models are trained on enormous amounts of text, and that includes every well-known framework. You can use that. If you frame your request inside something the model already understands, it doesn’t have to reconstruct the structure from your description, so it can spend its effort on the parts that are specific to you.

Ask for a business analysis and mention SWOT, and the model already knows the shape. It goes straight to applying it to your situation.

A worked example

Say you want help drafting a post about sustainable fashion:

Context: A fashion blogger writing about sustainable fashion for an
         eco-lifestyle site.

Goals:
1. Explain the environmental impact of fast fashion.
2. Give practical tips for building a sustainable wardrobe.
3. Highlight innovative sustainable brands.

Audience: Environmentally conscious millennials interested in fashion.
Style & Tone: Conversational and informative. British English.

Rules:
- Avoid jargon.
- Include at least three actionable tips.
- Mention no more than five brands.

Additional Information:
- Use the "4 Rs" framework (Reduce, Reuse, Recycle, Reimagine) as the
  main sections.
- Include a short section on circular fashion.

Now the model knows who it’s writing for, what you want out of it, and, through the “4 Rs”, a structure it already recognises and can hang the content on.

The short version

Effective prompting is mostly clear communication. Structure the input, borrow a framework the model knows, and you make its job easier and your output better. It can’t read your mind, so the clearer and more structured you are, the closer you get on the first try.