Six prompt moves that change the output noticeably

The Prompt Trinity gives you a framework for structuring your requests. Now let us look at six specific techniques you can use within that structure to push the quality of AI responses significantly higher. Think of these as power moves — each one is simple to apply and makes a measurable difference.

Technique 1: Give Rich Context

AI does not know anything about your situation unless you tell it. It does not know your company, your audience, your constraints, your preferences, or your goals. Every piece of relevant context you provide helps the AI narrow down from millions of possible responses to the one that actually fits your needs.

Rich context means sharing background information, purpose, audience details, and any constraints or requirements that matter.

Before:

Write a welcome email for new customers.

After:

Write a welcome email for new customers of FreshBox, a subscription meal-kit delivery service targeting busy professionals aged 25–40 in the UK. The email should arrive immediately after their first order. Our brand voice is warm, slightly playful, and health-conscious. The email needs to include next steps (when their box arrives, how to access recipes online, and how to contact support). Keep it under 250 words — our data shows mobile open rates drop sharply after that length.

The "before" prompt could be for any business in any industry. The "after" prompt gives the AI enough context to write something that genuinely sounds like it came from the FreshBox marketing team.

When to use this: Every time. There is almost no situation where additional context makes things worse. When in doubt, share more rather than less.

Technique 2: Show Examples

This is what experienced AI users call the "show, don't tell" principle. Instead of describing what you want in the abstract, you show the AI a concrete example of the style, format, or quality you are after.

Humans do this naturally. If you asked a colleague to write a report and they had never seen one of your company's reports before, you would hand them a previous one and say, "Like this." The same principle works brilliantly with AI.

Before:

Write product descriptions for our candles.

After:

Write product descriptions for three of our candles. Here is an example of our existing product description style:

Midnight Garden — Close your eyes and you are somewhere between a greenhouse and a library. Jasmine and old paper mingle with a whisper of damp earth. Burns for 45 hours. Hand-poured in small batches in Bristol.

Now write descriptions in this same style for:

1. A citrus and sea salt candle called "Morning Tide"
2. A cinnamon and vanilla candle called "Hearthstone"
3. A pine and woodsmoke candle called "Northern Trail"

By providing one example, you have communicated more about your desired tone, length, structure, and brand voice than any amount of abstract description could achieve. The AI will match the pattern — the sensory language, the evocative imagery, the practical details at the end.

When to use this: Whenever you have a specific style or format in mind, and especially when your desired output has a distinctive voice that would be hard to describe in words.

Technique 3: Set Constraints

Constraints are not limitations — they are creative guardrails. They tell the AI what to include, what to exclude, and what boundaries to respect. Without constraints, AI tends to produce output that is competent but generic, often longer than necessary and covering ground you did not ask for.

Useful constraints include word count, format requirements, tone restrictions, topics to avoid, and mandatory elements.

Before:

Explain blockchain to me.

After:

Explain blockchain technology in exactly three paragraphs. The first paragraph should use a physical-world analogy (not the "digital ledger" cliché). The second should explain why it matters for everyday people, not just technologists. The third should honestly acknowledge one significant limitation. Use no jargon without defining it. Aim for 200 words total.

The constraints in the "after" version do something powerful: they force the AI to be selective and thoughtful rather than producing a sprawling, encyclopaedic answer. You get a response that is focused, original (no "digital ledger" cliché), balanced (acknowledges a limitation), and appropriately sized.

When to use this: When you have a clear picture of the output dimensions, when you have received bloated responses in the past, or when you want to push the AI away from its default tendencies.

Technique 4: Break It Down

When you give AI a complex, multi-part task in one breath, it often handles some parts well and botches others. The solution is decomposition: break the task into numbered steps and ask the AI to work through them in sequence.

This works because it mirrors how AI processes information. By giving it a clear sequence, you ensure that each step gets full attention rather than being rushed to fit everything into one response.

Before:

Help me prepare for my job interview at a tech startup.

After:

Help me prepare for a product manager interview at a Series B fintech startup. Work through these steps in order:

1. Based on common Series B fintech priorities, list the five most likely interview topics
2. For each topic, write one strong example answer using the STAR method (Situation, Task, Action, Result), drawing on typical product management experience
3. List three insightful questions I could ask the interviewer that demonstrate strategic thinking
4. Identify two potential red flags or tricky questions I should prepare for, and suggest how to handle each

Each numbered step is clear, specific, and builds on the previous one. The AI works through them methodically, producing a comprehensive interview prep guide that covers multiple dimensions.

When to use this: For any task that involves more than one distinct component, for complex analysis, for tasks where the order of steps matters, or when you have previously received responses that felt incomplete or jumbled.

Technique 5: Ask AI to Think First

This technique is deceptively simple and remarkably effective. Before asking the AI to produce its final answer, you ask it to think through the problem, consider different angles, or outline its approach. This produces better reasoning and more nuanced output.

Why does this work? AI generates text sequentially — each word influences the next. When you ask it to reason before concluding, the reasoning process shapes and improves the final answer. It is like the difference between asking someone to blurt out an answer versus asking them to think for thirty seconds first.

Before:

What pricing strategy should I use for my online course?

After:

I am launching an online course on data visualisation for marketing professionals. Before recommending a pricing strategy, I would like you to think through the following:

• What are the main pricing models available for online courses, and what are the pros and cons of each?
• What factors specific to my niche (data visualisation for marketers) should influence the pricing decision?
• What are the common mistakes course creators make with pricing?

After thinking through these points, recommend a specific pricing strategy with your reasoning.

The "before" prompt gets you a quick answer. The "after" prompt gets you a well-reasoned recommendation backed by visible analysis. You can see the AI's thinking, which means you can evaluate whether the reasoning is sound — a critical skill we will develop further in Module 5.

When to use this: For any decision-making task, for complex analysis, for situations where you want to understand the reasoning (not just the conclusion), and when accuracy matters more than speed.

Technique 6: Ask AI to Help You Ask

This is the meta-technique — the one that makes all the others easier. When you are not sure how to phrase a prompt, or when you are tackling something unfamiliar, you can ask the AI to help you write a better prompt.

This is not a sign of weakness. It is a sign of sophistication. You are using the AI's knowledge of what makes a good prompt to improve your own communication.

Before:

Help me with my business plan.

After:

I am a first-time founder building a B2B SaaS product for small accounting firms. I need help developing my business plan, but I am not sure what to focus on first or what information you would need from me to be most helpful. Can you ask me 5–8 targeted questions that will help you understand my situation, and then suggest how we should structure our conversation to build out the plan step by step?

Instead of fumbling through a vague prompt, you have asked the AI to co-design the conversation. It will ask you smart questions, surface considerations you had not thought of, and propose a structured approach. You end up with a much better starting point than you would have reached on your own.

When to use this: When you are working in an unfamiliar domain, when the task is complex and you are not sure where to start, when you want to discover what you do not know, or when you simply want to get the most out of a conversation without spending ages crafting the perfect prompt.