CRIT, a four-part prompt worth stealing

In short

I picked this one up from Geoff Woods on Codie Sanchez’s BigDeal podcast. It is a two hour listen and I would not sit through all of it again, but the advice underneath is genuinely good, simple enough to explain in a minute and useful whether you opened ChatGPT for the first time last week or have been living in it for a year.

Four parts, and they spell CRIT. Context, Role, Interview, Task. The shape matters as much as the parts, because nearly all of a good prompt is context and the rest is roughly even.

C Context

Give the model everything about what you are doing, and the moment you think that is enough, give it more. This is where nearly all the quality comes from and it is the part everyone shortchanges. I dictate it rather than type it, because talking pulls far more out of your head than typing does and you stop quietly editing yourself down to three tidy sentences.

R Role

Tell it who you would actually want advice from, and get specific, niche even. Not a business consultant, but a strategic consultant who lives in contracts and retainers.

One thing the podcast did not say, but I would. Do not reach for you are a world-class expert. Controlled studies say it does not make the answer more accurate and can make it worse, all while making it sound more sure of itself. You are not puffing the model up, you are pointing its thinking.

I Interview

Before it answers, get it to interview you, one question at a time so each answer can shape the next. The podcast suggested defining three to five questions. I just leave it open and let it run until it has what it needs. This is the step that turns a decent prompt into a great one, because it fills the gaps you did not know were there.

T Task

Then, and only then, the actual job. What does the finished thing need to be? Keep it to the outcome, you have already done the heavy lifting above.

Context: [everything about what you are doing,
dictate this, do not type it]

Role: act as [the specific person you would want
advice from, not "an expert"]

Interview: before you answer, interview me one
question at a time until you have what you need

Task: [the finished thing you want]

A real example, the retainer

I used exactly this to draft a retainer for a company I was leaving, so they could still get my help after I went.

• Context. I dictated for ages, everything I had been doing, where I thought they would get real value, how I wanted it to work.
• Role. A strategic business consultant who specialises in contract and retainer work, not a generic business guru.
• Interview. I left it open. It asked about twelve questions, from the shape of the business to what I wanted out of the arrangement, all sharp.
• Task. Write the retainer.

The draft that came out was miles better than anything I would have typed cold.

The bit I would not skip, review

Once I had the draft, I sent it back through three separate review passes, each one told to act as a critic and hunt for what was missing rather than tidy the wording. They caught things I would have happily signed. There was no minimum spend on the ad-hoc extra time, and I had quietly agreed to cover my own travel.

That is the difference between a prompt and a process, and it is worth knowing why it is not a free pass. A model reviewing its own draft is still working with the same brain that wrote it, which is the point made in what thinking actually means. The review catches gaps and inconsistencies well. It does not make the document correct, and you still read it before it goes anywhere.

That is it

If you are reading this thinking you are past frameworks, honestly, so was I. I do not think in frameworks when I am working and I do not need to. But every time I go back to having it interview me before it writes anything, I am surprised by how good the questions are.

Context, Role, Interview, Task, and a review to finish. If you try it, tell me what you got out of it, I read everything.