Prompt Crafting: How to Get Better Answers From AI

People try AI, get a bland or generic answer, and conclude the tool isn't very good. Usually the tool is fine — the prompt was thin. "Write me a marketing email" gives a forgettable email because you've told it almost nothing: no audience, no product, no tone, no goal. The model isn't a mind reader; it fills the gaps you leave with the most average, middle-of-the-road guess. Garbage in, generic out.

Learning how to write better AI prompts is mostly learning to brief well — the same skill as giving a clear instruction to a colleague. The difference between a useless answer and a great one is almost always how much useful direction you put in.

How do I write better prompts for ChatGPT or Claude?

Think of a prompt as a brief, and remember that a good brief carries four things. Context: the situation, the goal, and who it's for — "this is a follow-up email to a prospect who went quiet after a demo." Role: who you want the AI to act as — "act as an experienced sales rep." Format: exactly what you want back — "give me three short versions, each under 80 words, casual tone." And examples or constraints: a sample of the style you like, or hard limits — "match the tone of this email I'll paste; don't be pushy."

The other half of the skill is iteration. Your first prompt rarely lands perfectly, and that's fine — the move is to refine, not restart. "Make it shorter," "more formal," "focus on the second option," "you missed the deadline detail." Treating it as a conversation, steering with each reply, is what separates people who get great results from people who give up after one mediocre try.

How to craft a better prompt, step by step (about 10 minutes)

You need an AI tool and one real task you want help with.

1. State the context and goal first. What's the situation, what are you trying to achieve, and who's it for? This single step fixes most weak prompts on its own.
2. Assign a role. "Act as a copywriter / analyst / careful editor." A role focuses the model's tone and priorities toward what you actually want.
3. Specify the exact format. Length, structure, number of options, tone. "Three bullet options, one sentence each" beats hoping it guesses your format.
4. Add an example or constraints. Paste a sample you like, or set limits — "avoid jargon," "keep it under 100 words." Examples are the fastest way to convey a style.
5. Run it, then refine — don't restart. Read the output and steer: shorter, sharper, fix the thing it missed. Two or three rounds of nudging usually gets you there.
6. Save prompts that work. When a prompt nails a recurring task, keep it. A small personal library of good prompts compounds over time.

A worked example

Compare two prompts. Weak: "Write a LinkedIn post about our new feature." It returns something generic and forgettable. Strong: "Act as a B2B marketer. Write a LinkedIn post announcing our new scheduling feature for busy small-business owners. Goal: get them to try it. Keep it under 120 words, conversational, lead with the pain of double-booking, end with one clear call to action. Avoid buzzwords." The second returns something close to usable on the first try — and a quick "make the opening punchier" gets it the rest of the way. Same tool, vastly different result, entirely because of the brief.

When this is most useful

Strong prompting matters most for anything where quality and specificity count — drafting content, analysis, structured outputs, or any task you'd otherwise have to heavily rewrite. The payoff grows with how often you do the task, since good prompts are reusable. It matters less for trivial one-line questions, where a quick ask is fine. And remember: a better prompt improves relevance and shape, not factual reliability — you still verify anything important, because a well-formatted answer can still be confidently wrong.

The takeaway

Bad AI answers usually come from thin prompts, not bad tools. Brief the AI like you'd brief a sharp colleague: give context and goal, assign a role, specify the format, and add an example or constraints. Then refine through a couple of rounds instead of starting over, and save the prompts that work. Master this and the same tool that gave you generic mush starts giving you genuinely useful first drafts.

This is one of Funstorming's 100 quests — bite-sized soft skills methods you actually put into practice, not just read about. Try it, then bring your result (or your sticking point) to the Funstorming community of practice (CoP), FunHub | Your Soft Skills Playground.

#funstorming #softskills