Your 5-Minute Guide to AI Prompting (No Fear Required)

Hey there,

So you’ve probably tried ChatGPT or Claude by now. Maybe you asked it to write some SQL and got something that sort of worked. Or maybe you’re still staring at that blank prompt box wondering what exactly you’re supposed to type.

Either way, you’re not alone.

But what we fail to recognize sometimes is:
​Good prompting is just good communication.

That’s it. No magic formulas. No secret techniques.

You already know how to explain what you need to a colleague or document your analysis process. Same skills, different audience.

Here's What's Really Going On

Most prompting advice you'll see online isn't wrong, but it can be overwhelming when you're just starting out.

"Use few-shot learning!"
"Implement chain-of-thought reasoning!"
"Engineer your prompts!"

Those techniques work.
But if you're new to AI tools, they can make the whole thing feel impossibly complicated.

Here's the thing:
you don't need to master advanced techniques to get value from AI.

For most day-to-day data tasks, you just need to be clear about what you want.

Start simple.
Get comfortable.
The fancy stuff can come later.

What Actually Works

I use AI tools daily now - for SQL debugging, chart reviews, explaining analysis to stakeholders. Here’s what I’ve learned:

1. Give Context (Because AI Doesn’t Read Your Mind)

Think about it this way: if someone walked up to you and said “Cook food,” what would you do?

You’d ask questions, right?

• Cook for how many people?
• What occasion?
• Any dietary restrictions?
• What ingredients do you have?

Without context, even simple requests become impossible to fulfill properly.

AI has the same problem, just faster and more confidently wrong.

Instead of:

Try:

The first version could mean anything.

• Maybe the syntax is wrong.
• Maybe it’s running slow.
• Maybe the logic is off.
• AI will guess, and probably guess wrong.

The second version tells AI exactly what success looks like and what’s currently broken. Now it can actually help.

Here’s what happens without context:
You get generic advice that doesn’t fit your situation. AI will give you textbook answers for problems you’re not actually having.

With context? You get targeted help that actually works.

2. Be Specific About What You Need

Don’t make the AI guess what you’re after:

• “Explain this in simple terms”
• “Find errors in my logic”
• “Suggest three ways to visualize this data”
• “Help me clean up this messy query”

The more specific you are about the deliverable, the better the response.

3. Follow Up When Needed

The first response often isn’t perfect. That’s fine.

“Make it simpler” or “Actually, I need more detail on the forecasting part” or “This doesn’t account for our data structure.”

It’s a conversation, not a one-shot deal.

A Simple Template That Works

Once you get the hang of providing context and being specific, here’s a simple starting structure I use for many task-related requests:

Example:

That covers all the context AI needs without overthinking it.

When It Doesn’t Work

Even with good context and clear requests, sometimes AI gives you wrong answers.

Confidently wrong answers.

When that happens, ask it to double-check:

or

AI tools are great at catching things you missed.

They’re terrible at business context and knowing when rules have exceptions.

Always review and sanity-check the output.

The Bottom Line

You already have the communication skills for this. You know how to explain problems, provide context, and ask follow-up questions.

AI prompting is just using those skills with a new tool.

The difference between people who get value from AI and those who don’t? It’s usually about approach, not technical skill.

Try This Today

Pick something small you’re working on:
- A SQL query that’s not quite right
- A chart that needs improvement
- Explaining your analysis to someone non-technical

Open ChatGPT or Claude.
Be specific about what you need help with.
See what happens.

Don’t overthink it.
Just try it.

Learning to use AI effectively is just another skill in your toolkit.

And honestly, you probably have better instincts for this than you think.

Chat soon,
Donabel

P.S. Speaking of AI prompting - I’m launching “Build to Sell: 7 Days to Your First Digital Data Product” on July 24th.
​
AI prompting is actually a core piece of the course, and you’ll get a complete prompt library specifically for data professionals wanting to build their own digital data product.
​
Plus some bonuses I’m excited about:

• a Digital Data Product Ideas Guide, and
• ​a full year access to ALL articles and prompts in “Teach Data with AI” newsletter (teachdatawithai.substack.com)

​
The newsletter alone is a game-changer for people learning to use AI effectively in data education and training. Teach Data with AI has 2-3 new posts per week, many of them with accompanying full prompts (for example, how to explain 95% confident in plain English)
​
​
Here's a sneak peak at the course content and the accompanying AI Prompt Library:
​

​

​

​