Stop Treating AI Like Google

How to use AI correctly - 4 elements of a good prompt.

Most people use AI the same way they use a search engine. Type a question, read the answer, move on. And then they're disappointed by what comes back, because the answer is generic, vague, or sounds like it was written for everyone and nobody.

That isn't AI failing. That's AI being used wrong.

A search engine's job is to find something that already exists. AI's job is to produce something that doesn't exist yet, based on what you tell it. The difference is enormous. If you give it almost nothing, it has almost nothing to work with, and the output reflects that.

The shift is small but changes everything: brief it, don't query it.

The how-to

Every time you ask AI to do something, give it four things before you let it start.

  1. A role - Who should it be while answering? A school admissions manager? A copywriter? A parent reading this email for the first time? Role shapes everything that follows.

  2. The input - What's it working with? Your existing welcome letter. Last week's enquiry figures. A competitor's open day invitation. The more specific, the better.

  3. The task - What exactly do you want it to do? Summarise? Compare? Pull out themes? Draft three options?

  4. The output format - How should the answer come back? Three bullet points? A 200-word paragraph in a warm tone? A table with three columns?

These four elements turn a vague question into a proper brief. The output will improve immediately, often dramatically, on the very first try.

Example prompt

Instead of typing:

"What does my enquiry data mean?"

Try:

"You are a senior admissions analyst. Below is a CSV export of our enquiry data from the last 12 months, broken down by source, year group, and month. I need you to identify the three most significant patterns in this data, flag any sources that appear to be declining year-on-year, and tell me the one question I should bring to my SLT meeting. Return it as a short bullet-point list, under 200 words, in plain English. Here is the data: [paste or attach]."

Same task. Completely different output.

The shift

Most people are closer to getting the hang of AI than they realise.

The only thing standing between a generic output and a useful one is the briefing. You already write briefs at work, for designers, for photographers, for your own team.

This is the same skill, applied to a new tool.

Simple.

Previous
Previous

Website Audit Prompt