Prompting is a skill. That sounds obvious, but most people don't treat it that way. They type a question, get an answer they don't love, and assume AI just isn't very good at that thing — when really, the problem is in how they asked.
The good news: a handful of habits can dramatically improve the quality of what you get back. Here are five worth building.
Habit 1: Be specific about what you actually want
Vague prompts produce vague answers. If you ask "write me an email," you'll get a generic template. If you ask "write a short follow-up email to a client who hasn't responded to a proposal after two weeks — keep it warm but direct, and give them an easy way to say they're not interested," you'll get something you can actually use.
The more specific you are about the output you want — the length, the tone, the purpose, the audience — the less work you'll need to do afterwards. Every vague word in your prompt is a guess the model has to make on your behalf.
Habit 2: Give context before giving the instruction
Think about how you'd brief a colleague. You wouldn't just say "summarise this" — you'd explain what it's for, who's going to read it, and what level of detail they need. AI works the same way.
A prompt like "I'm preparing a briefing for a non-technical manager. Summarise the following report in plain language, focusing on business impact rather than technical details" will outperform "summarise this" every single time. Context shapes the response before the instruction even lands.
Think about how you'd brief a smart colleague who just joined the team. That's how you should brief an AI.
Habit 3: Specify the format you need
AI will default to whatever format seems most natural for the task. Sometimes that's a long essay when you needed three bullet points. Sometimes it's bullet points when you needed flowing prose.
Tell it what you want. "Give me your answer as a numbered list," or "Keep this to two paragraphs," or "Format this as a table with three columns." This isn't about being prescriptive for the sake of it — it's about getting output that actually fits where it's going without you having to reformat it manually.
Habit 4: Iterate, don't restart
Most people abandon a conversation if the first response isn't right. Better prompting means staying in the conversation and directing the model towards what you need.
"That's good but the tone is too formal — make it more conversational." Or "The first point is strong but the third one is too generic — can you give me something more specific?" Or simply: "Try again with more detail on the second section."
AI handles feedback well. You don't have to start a new chat — you can just tell it what to fix. This iterative approach is how good outputs actually happen.
Habit 5: Save your best prompts
When you find a prompt structure that works well for a recurring task — drafting updates, summarising documents, researching topics, preparing talking points — write it down somewhere. You've just built a reusable tool.
The best prompt libraries aren't fancy. A simple document or folder of saved prompts can save you significant time over the course of a week. And when a new AI tool arrives, your prompts often transfer with minimal adjustment.
The underlying skill
All five of these habits share a common root: they're about being clear and deliberate in how you communicate what you need. That's a skill that applies across tools, across models, and across whatever changes happen in the AI landscape over the next few years.
The best way to build it is to practise on real tasks — not hypothetical exercises, but the actual work you do every day. That friction is where the learning happens.
Build the skill properly
Prompting for Work — a CertifAI course.
Go beyond the basics. Guided lessons, real practice tasks, and AI Tutor feedback to help you develop prompting habits that actually stick.
Explore Prompting for Work →