Prompting 101: How to Ask AI for What You Actually Want

Why vague prompts get vague answers

A prompt is just what you type to AI: the request plus any information you give it. When the request is vague, AI doesn't stop to ask what you meant. It fills every gap with the most generic answer that sounds right.

Think of AI as a brilliant new hire on day one. It writes well, but it knows nothing about your company, your buyer or yesterday's call until you tell it. So when an answer comes back bland, the fix is usually a better ask, not a better tool. Most bad output is a context problem, not a model problem. How AI Assistants Actually Work explains why.

In sales, generic means ignored. You don't need clever wording or secret phrases (that idea is myth 4 in 10 Myths About AI in Sales). You need a short checklist.

The five parts of a good prompt

The Wicked Growth Club course teaches one framework for every prompt, called GCIRF. Each letter is a question to answer before you hit enter.

You don't need all five every time. A quick question ("What does a VP of RevOps usually own?") can be one line. The more the answer matters, the more parts you fill in. Anything a buyer will read deserves all five.

Before and after

Here's one real task done both ways. Sam is a prospecting rep at Ramplight, a fictional platform that helps sales teams onboard new reps faster. Sam has a first call tomorrow with Priya Raman, VP of Sales at Calder Freight Systems, also fictional.

Help me prep for my call with Calder Freight tomorrow.

Goal: help me run a 20 minute first call with Priya Raman, VP of Sales at Calder Freight Systems, that ends with an agreed next step.

Context: I'm a prospecting rep at Ramplight. We help sales teams onboard new reps faster with AI coaching. Priya took the call after my email about their open prospecting roles. I don't quote pricing.

Inputs: my notes are pasted below. Use only these facts about Calder and Priya. If something important is missing, list it as a question for me instead of guessing.

Rules: plain language. No more than 5 discovery questions. Don't invent numbers or customer names.

Format: a one page prep sheet with 4 headings: what I know, what I need to learn, my 5 questions, and a suggested next step to propose.

<my notes>
[paste your notes here]
</my notes>

Notice that the "after" prompt isn't clever. It's just complete. Every line answers one of the five questions.

Run through this before you send any prompt that matters.

The iteration habit

People who get great results from AI share one habit: they don't start over. The first answer is a draft. You make it great with feedback, the way you'd coach a new rep.

Weak feedback sounds like "make it better." AI has to guess what "better" means, so it guesses generically. Strong feedback follows a simple recipe: what to keep, what to change and by how much.

Two or three rounds of specific feedback usually beat rewriting your prompt from scratch. If a chat goes badly off track after about four rounds, open a new chat and start with a better prompt. Long chats can drift as they fill up the context window, the amount of text AI can consider at once.

Three beginner techniques

The course teaches seven prompting techniques. These three are the easiest place to start, and each one works in your company approved AI assistant.

1. Show examples. Adjectives like "professional but friendly" mean something different to everyone. Two or three real examples show AI exactly what you mean. (The technical name for this is few shot prompting.)

Write a follow up email to Priya Raman at Calder Freight Systems after today's first call. Match the style of the two follow up emails I wrote below: same length, same tone, same way of ending.

<my emails>
[paste 2 of your best follow up emails]
</my emails>

2. Ask AI to ask you questions first. When you're not sure what context to give, let AI tell you what's missing.

I need to write a follow up email to Priya Raman, VP of Sales at Calder Freight Systems, after our first call. Before you write anything, ask me up to 5 questions about what you need to know. Wait for my answers.

3. Ask for options. One answer gives you a yes or no choice. Three answers give you something to compare, pick from and combine.

Give me 3 different ways to open my call with Priya Raman at Calder Freight Systems: one that starts with their open prospecting roles, one that starts with a question about how new reps ramp today, and one that simply confirms the agenda. One or two sentences each.

When a prompt works, save it. Swap the specific details for blanks in curly braces, and it becomes a prompt template you can reuse on every account.

Goal: help me run a {length} first call with {name}, {title} at {company}, that ends with an agreed next step.
Context: use my saved context. {how the meeting was booked}.
Inputs: my notes are below. Use only these facts. List missing information as questions for me.
Rules: plain language. No more than 5 discovery questions. Don't invent numbers or customer names.
Format: a one page prep sheet: what I know, what I need to learn, my questions, a next step to propose.

{my notes}

If you save a longer note about your company, product and buyers once, you won't need to retype the context line. That note is called a seller context file.

Ready for more? Once GCIRF feels natural, AI Prompts for Sales: 15 Tested Prompts gives you ready prompts for research, outreach, calls and pipeline, all built the same way. For outreach in particular, read How to Write Cold Emails with AI That Don't Sound Like AI.

Quick check

FAQ