How To Write Prompts That Actually Work

Most people talk to AI like it’s a vending machine. Here’s how to treat it

like the powerful thinking partner, it actually is — and get a lot

better results every time.

You’ve typed a question into an AI. You got back something technically

correct but completely useless — too generic, too long, or totally

off-target. You’ve stared at the response, frustrated, and thought: this

thing is supposed to be intelligent? The problem wasn’t the AI. It was

the prompt.

Prompt engineering sounds intimidating — like something only developers or

researchers do. In reality, it’s just the art of communicating clearly with

an AI system. Once you understand a handful of core principles, you’ll

notice results improve immediately by approximately ten times.

WHY MOST PROMPTS FAIL?

AI models are pattern-completion engines trained on an ocean of human text.

They’re remarkably capable — but they can only work with what you give them.

Vague inputs produce vague outputs. If you ask a chef to “make something

good,” don’t be surprised when the result isn’t quite what you had in mind.

The same principle applies here.

“The AI isn’t guessing what you need — it’s completing the pattern you

started. Make that pattern precise.”

Most bad prompts share three failure modes: they lack context, they’re

ambiguous about format, and they give the model no constraint on length or

depth. The fix for all three is the same — be specific.

THE FOUR ELEMENTS OF A GREAT PROMPT

Think of any strong prompt as having four components. You don’t always need

all four, but knowing them makes the difference between an average result

and an exceptional one.

1. ROLE

   Tell the AI what hat to wear. “Act as a senior copywriter” or “You are

   a friendly financial advisor” primes the model to draw from a specific

   domain of knowledge and tone.

2. TASK

   State precisely what you want. Not “write something about marketing”

   but “write a 200-word LinkedIn post announcing our product launch,

   aimed at B2B founders.”

3. CONTEXT

   Give relevant background. Who is the audience? What’s the purpose?

   What constraints exist? More context = tighter output.

4. FORMAT

   Specify the output shape. “Give me a bullet list,” “Write in plain

   paragraphs,” “Use a table,” or “Keep it under 150 words” all

   dramatically change the result.

SEE IT IN ACTION

Abstract advice is easy to ignore. Here’s the same request, written

two ways:

WEAK PROMPT:

“Write an email about the project delay.”

STRONG PROMPT:

“Write a professional but warm email from a project manager to a client,

explaining that our product launch is delayed by two weeks due to a QA

issue. Acknowledge the inconvenience, provide a revised timeline, and end

with confidence. Keep it under 150 words.”

The second prompt specifies the sender, the audience, the reason, the tone,

the action, and the length. The AI now has everything it needs — and it will

produce something you can use almost immediately.

ONE TRICK THAT CHANGES EVERYTHING: SHOW, DON’T TELL

One of the most underused techniques is giving the AI an example of what

“good” looks like. Instead of describing the tone you want, paste a sentence

or two that captures it. Instead of saying “formal but approachable,” show

the model a line that achieves that. Examples anchor the model’s output far

more effectively than adjectives ever will.

This technique — called few-shot prompting — works even with a single

example. Try: “Here’s a sample of my writing style: [paste a paragraph].

Now write an intro for my newsletter in the same voice.” The results are

almost always noticeably better.

WHEN RESULTS STILL DISAPPOINT

Even good prompts sometimes produce mediocre first drafts. The best approach

is to iterate — not start over. Respond with what’s wrong: “This is too

formal — make it conversational” or “Condense the third paragraph by half.”

AI models are designed for dialogue. Each follow-up refines the output

without losing the thread of what you’ve already built together.

Think of it less like issuing a command and more like collaborating with a

talented but literal-minded colleague. One who never gets tired, never

judges your ideas, and genuinely benefits from your feedback.

“Prompting is a conversation, not a command. The best results come from

iteration, not perfection on the first try.”

FIVE RULES TO LIVE BY

1. Be specific. Every vague word in your prompt is an invitation for the

   AI to guess — and it may guess wrong.

2. Set the format upfront. Tell the model what shape the output should

   take before it starts writing.

3. Give examples when you can. A sample is worth a hundred adjectives.

4. Add constraints. Word limits, tone restrictions, audience details —

   boundaries improve focus.

5. Iterate rather than restart. Build on what the model gives you; refine

   rather than rewrite from scratch each time.

THE BIGGER PICTURE

Prompt engineering isn’t a niche technical skill — it’s rapidly becoming a

fundamental form of literacy. As AI becomes embedded in how we work, learn,

and create, the ability to communicate clearly with these systems will be as

valuable as knowing how to write a clear email or run a decent search query.

The good news: you don’t need a computer science degree, a special

certification, or weeks of study. You need to slow down slightly before you

type, think about what you actually want, and give the model enough to work

with. That’s it. Everything else is practice.

Start with your next prompt. Try the framework. Notice the difference.

Then do it again.

Scroll to Top