How Do I Write Prompts That Actually Get Good Results?

By

·

How Do I Write Prompts That Actually Get Good Results?

How Do I Write Prompts That Actually Get Good Results?

The secret to better AI prompts isn’t clever wording — it’s structure. Add context about your situation, tell the AI what format you want, and drop in a quick example. Do those three things consistently and your results will improve right away, no technical skill needed.

Why this comes up

You’ve typed something into an AI tool and gotten back a wall of generic text that missed the point entirely. The easy conclusion is that the tool is broken or you’re doing something wrong. You’re not — you’re just missing a few simple habits that nobody teaches beginners upfront.

The frustrating part: most AI guides assume you already know what a “prompt” is supposed to look like. This one doesn’t. These are the three habits that actually move the needle for normal people doing normal tasks.

The honest answer

Good prompts aren’t magic spells. They’re just clear communication. Think of the AI like a capable new colleague who doesn’t know your situation yet — the more you tell them, the better their first draft will be.

Here are the three habits that make the biggest difference:

Habit 1 — Give context (who you are, what you need, why)

Most people write what they want. Strong prompts also explain the situation.

  • Weak: “Write an email to my boss.”
  • Strong: “I’m a junior designer asking my manager for one extra week on a project because the client added last-minute revisions. Write a short, professional email.”

Context tells the AI what kind of answer actually fits your life. Here are a few more examples of how it changes things:

Without context With context
“Explain compound interest” “Explain compound interest like I’m 14 and haven’t done maths since primary school”
“Write a cover letter” “I’m applying for a part-time barista job. I have no café experience but I’ve worked in retail for 2 years. Write a short cover letter.”
“Summarise this document” “I’m a nurse. Summarise this research paper in plain English so I can share the key finding with a patient.”
“Give me workout ideas” “I have 20 minutes, no equipment, and a bad knee. Give me a low-impact home workout.”

In every case, the extra sentence transforms a generic answer into something usable.

Habit 2 — Specify a format

The AI will guess at structure if you don’t give one. Sometimes it guesses wrong and produces a wall of text when you needed a list, or three paragraphs when you needed a one-liner.

  • Say things like: “in bullet points,” “under 100 words,” “as a numbered list,” “in a friendly tone,” or “write it as a table.”
  • Combine them: “Give me five bullet points, plain language, no jargon” is a complete format instruction.

This one habit alone cuts vague, overlong responses dramatically. Some format instructions that work well:

  • Length: “Under 50 words” / “About 200 words” / “One paragraph only”
  • Structure: “As a numbered step-by-step” / “As a table” / “As a comparison”
  • Tone: “Casual and friendly” / “Formal and professional” / “Like you’re explaining to a 10-year-old”
  • Angle: “Focus only on the downsides” / “Include one real example” / “End with a specific action I can take today”

Habit 3 — Show an example (or describe one)

This is the most underused tip for beginners. If you have a sample of what good looks like, paste it in or describe it.

  • “Here’s an email I liked the tone of: [paste it]. Match that style.”
  • “Write it the way a friendly pharmacist would explain it to a patient.”
  • “Here’s a post I liked: [paste]. Write something similar for my situation.”

Examples anchor the AI to your standard, not a generic one. You don’t need a perfect example — even “something like this, but shorter and less formal” gives the AI a useful anchor.

The upgrade formula: Who you are + what you need + what format + one example = a prompt that actually works.


Before & After: 5 Real Rewrites

Here’s what the habit upgrade looks like in practice across different tasks.

1. Writing an email
– ❌ Before: “Write an apology email.”
– ✅ After: “I forgot to send a report my colleague was waiting for. Write a short, sincere apology email that acknowledges the delay without over-explaining. Keep it under 80 words.”

2. Researching a topic
– ❌ Before: “Tell me about sleep.”
– ✅ After: “I sleep 6 hours a night and feel fine but read that I need 8. Give me a plain-English explanation of whether this is actually a problem, and if so what one thing I should try first.”

3. Getting a decision made
– ❌ Before: “Should I accept this job offer?”
– ✅ After: “I have a job offer that pays 15% more but requires 30 minutes of extra commuting each way. My current job is stable but not growing. List the pros and cons in a table and then give me your honest take.”

4. Learning something faster
– ❌ Before: “Explain photosynthesis.”
– ✅ After: “Explain photosynthesis using only a cooking analogy. I’m an adult who hasn’t studied science since school.”

5. Getting creative work done
– ❌ Before: “Write a birthday message.”
– ✅ After: “Write a warm but not cheesy birthday message for my colleague who’s turning 40. We’re friendly but not close. Two sentences max.”


The Follow-Up Habit (Iteration)

Even with a good prompt, the first response often needs one or two tweaks. The fastest fix isn’t rewriting your prompt from scratch — it’s a short follow-up instruction.

These phrases work in any AI tool:

  • “Make it shorter” — cuts bloated responses instantly
  • “Make it more casual / more formal” — adjusts tone
  • “Give me three alternatives” — useful when the first option is almost right
  • “Add a specific example” — when the answer is too abstract
  • “Explain the third point in more detail” — drill into one part without redoing everything
  • “Now write it from the other person’s perspective” — useful for tricky conversations

One good follow-up beats ten rewrites of the original prompt.


What to do

Do:
– ✅ Start every prompt with one sentence about your situation
– ✅ End every prompt with a format instruction (“bullet points,” “under 150 words,” etc.)
– ✅ Paste in an example when you have one — even a rough one helps
– ✅ If the first answer misses, add detail rather than starting over: “Make it shorter and more casual”
– ✅ Use follow-up messages to fine-tune — it’s faster than starting over

Don’t:
– ❌ Write one-line prompts for complex tasks and expect a polished result
– ❌ Accept the first response if it’s off — one quick follow-up fixes most problems
– ❌ Overthink the wording; a longer, messier prompt beats a polished but vague one
– ❌ Think you need to “crack the code” — these three habits cover 90% of everyday use


FAQ

Does this work the same way across different AI tools?
Yes. These habits apply whether you’re using ChatGPT, Gemini, Claude, or any other major assistant. Clear context, clear format, clear example is a universal principle — not platform-specific.

How long should a good prompt be?
Long enough to answer: who am I, what do I need, and what should it look like? That’s often two to four sentences. Specific always beats short and vague, but you don’t need an essay.

What if I don’t have an example to share?
Describe what good looks like instead. “Write it the way a friendly teacher would explain it to a curious ten-year-old” works just as well as a real example.

Should I say please and thank you to the AI?
You don’t need to — but it doesn’t hurt. Politeness has no mechanical effect on AI responses, but some people find that framing the prompt conversationally (“Could you help me…”) produces slightly friendlier tone in return. Do what feels natural.

How do I get it to stop being so formal and stiff?
Add a tone instruction directly: “Write this in a casual, conversational tone — like texting a knowledgeable friend.” Or show it an example with the tone you want and say “match this style.” Formal is the AI’s default; you just have to tell it otherwise.

What’s a “system prompt” — do I need to know about it?
Not for everyday use. A system prompt is a behind-the-scenes instruction that sets the AI’s behaviour before the conversation starts. It’s used by developers building apps, not by people using ChatGPT or Gemini day-to-day. You can safely ignore it until you’re building something.


Bottom line

Better AI results come from better setup, not better luck — give it context, a format, and an example, and useful answers become the rule rather than the exception. When the first response misses, a short follow-up almost always fixes it without starting over.

What prompt habit changed things for you — or what’s still tripping you up? Drop it in the comments. We read every one.

Related: ChatGPT vs Gemini vs Claude — which one should you actually start with?


You might also like:

⭐ Want every prompt in one place?

All 136 tested prompts from our daily skills are free on
GitHub — DailySkill Prompt Library.
Star it to get new prompts as we publish.


Share this

Share on X Facebook LinkedIn

Read next