Lesson 001Field guide · Prompting 101
What Makes a Good Prompt
The anatomy of prompts that get useful results — and how to fix ones that don't.
You'll learn
- Why vague prompts get vague results
- The four parts that do the heavy lifting: role, task, context, format
- How to spot what a weak prompt is missing
- How to upgrade a weak prompt step by step
- How to rewrite a real prompt yourself
01Section
Why most prompts disappoint
You've probably typed something like this into an AI tool and gotten a wall of generic text back:
A typical first prompt
Write something about marketing.
The tool didn't fail — the prompt did. AI can't read your mind: marketing for whom? What kind of something? How long? For what goal? Every detail you leave out, the AI fills in with a guess, and its guesses are aimed at the average of everyone — which is why the result feels like it was written for no one.
Key takeaway
AI fills every gap you leave with a guess. Good prompting is mostly gap-closing.
Checkpoint
02Section
The four parts that do the heavy lifting
If you took Intro to AI, you met these briefly. Either way, almost every strong prompt is built from the same four parts:
Plain-language definition
Context — the background the AI can't know unless you say it — who this is for, what the situation is, and what you're trying to achieve. It's the part beginners skip most, and the part that changes results most.
A fifth part — constraints — sets the boundaries: length limits, tone rules, things to avoid. Think of it as telling the AI what a wrong answer looks like.
Checkpoint
03Section
What a fix looks like
Here's the same request, before and after those parts are filled in:
Weak prompt
Write something about marketing.
Improved prompt
Act as an experienced marketer for small local businesses. Write three simple marketing ideas for a family-run bakery that wants more weekday morning customers. The audience is nearby regulars, not tourists. Give each idea as one short paragraph with a rough cost estimate. Plain language, no marketing jargon.
Same topic, same AI tool — the only thing that changed is how much the AI had to guess.
Give it a role: "Act as an experienced local marketer" sets the perspective and the standard the answer is written to.
Name the task precisely: "Write three marketing ideas" beats "write something."
Add the context: Who it's for, what the situation is, what you want to happen.
Describe the output: Format, length, structure — what should the answer look like?
Set the boundaries: Tone, things to avoid, limits. Optional, but it sharpens everything.
Why not just fix it with follow-up questions instead?
You can — iterating is a real skill and a later lesson covers it. But every follow-up spends your time on repairs a better first prompt would have avoided. Starting strong usually gets you 80% of the way in one step, and follow-ups become fine-tuning instead of rescue work.
Checkpoint
04Section
The mistakes to watch for
- Too vague — "help me with my website" gives the AI nothing to work with.
- Five asks in one prompt — split big requests into steps instead.
- No output format — you'll get whatever shape the AI feels like.
- Assuming it knows your business — it only knows what you tell it.
Privacy note: Adding context never means adding secrets. Describe your situation without pasting customer data, passwords, or anything you wouldn't put in an email to a stranger.
Pause and think: Think about the last prompt you typed into an AI tool. Which of the four parts was missing?
Checkpoint
Prompt rewrite exercise
Rewrite a weak prompt yourself
Below is a real-world weak prompt. First spot what it's missing, then build the improved version part by part — pick a preset or write your own. When it's assembled, copy it into your own AI tool and compare the result against the weak version.
The weak prompt
Write an email to my customers about the changes.
Quick check
3 quick questions — no pressure
There's no pass or fail here. Answer them all, and we'll show you the answers either way.