Via Dylan Davis

On data analysis: "I'm sure there's tons of you out there dropping data into ChatGPT or Claude and trying to have it assess that data and give you insights from it. […] Imagine we have a prompt asking the the AI to analyze our sales data and tell us what's happening. This is a basic prompt. There's a lot wrong with this. […] What's missing from this prompt and why is it wrong?

First thing you weren't clear on: the question that you wanted to ask. You just generally said, "Hey, tell me what's happening." It's good to be clear on what you want to get from the data. And if you don't know the answer to this question or you're not really sure what question you need to ask, you can at least state the decision you're trying to make from the data set. So you give AI data and say, "Hey, I need to make these types of decisions from this data. Can you pull insights out from it that can help me make that decision?"

— The AI Prompting Habit That Saves Me 10+ Hours a Week, at 8:36

---

The generic bad prompt, "Analyze my sales data and tell me what's happening," produces obvious, useless insights like "Sales increased 12% in Q3"—something you could see yourself.

What data analysis prompts must have:

• Specific questions: Be clear about what you want to learn from the data
• Decision context: If you don't know the business question to ask, at least state what decision you need to make
• Targets and quotas: What benchmark are you trying to hit?
• Definition of success: What does good look like, so the AI knows the boundaries

Doing all of the above lets you to state your intent. Always share what you plan to do with the insights after you get them. Sharing your intent allows the AI to figure out the rest for you. The AI can work backwards from your goal to determine what analysis matters.