← All guides
Business · Data & Insights ·Claude, ChatGPT

Turn Raw Customer Data Into Insights Your CEO Can Act On

Data is not the deliverable. A decision is. Here is the reusable prompt I use to turn a messy export into a one-paragraph insight leadership can act on.

Give me data, give me numbers, give me a good spreadsheet. That is my happy place. But here is what I learned leading a data team: the spreadsheet is never the deliverable. The decision is.

Nobody in the C-suite wants your export. They want to know what to do. Here is how I turn one into the other, fast.

The what, so-what, now-what

Every useful insight answers three questions in order:

  1. What happened in the data.
  2. So what it means for the business.
  3. Now what we should do about it.

Most reports stop at “what.” That is a data dump, not an insight. The value you add as a leader is the “so what” and the “now what.”

The prompt

Paste your data with this. It forces the structure that turns numbers into a decision.

You are a sharp marketing analyst. I will paste customer data. Write a
short insight memo in three parts: WHAT (the one or two things that stand
out in the data), SO WHAT (what it means for the business, in plain
English), and NOW WHAT (one specific recommendation). Show which numbers
you based this on. Flag anything you are unsure about. No jargon.

You get a memo a CEO could read in a minute, grounded in the actual numbers, with a recommendation attached.

The tell of a good insight

If someone can read it and immediately know what to do differently, it is an insight. If they finish it and say “interesting,” you stopped at “what.” Push it one more step every time.

This same move works on your own life, by the way. Paste your spending, your sleep data, your time logs. What, so what, now what. It is the same skill, pointed at a different dashboard.

Common questions

I'm not a data analyst. Can I still use this?

Yes. That is the point. You do not need to write formulas. You need to ask the right question and sanity-check the answer. The prompt handles the structure; you bring the business context.

How do I trust the output?

Treat it as a smart first pass, not gospel. Always ask it to show which numbers it used and to flag anything uncertain. Then check the one or two claims your decision actually rests on.

Is it safe to paste real customer data into an AI tool?

Use your company's approved tool and follow your data policy. If the export has names, emails, or anything personal, strip those columns first. The prompt only needs the numbers and the patterns, not who the customers are.

The Newsletter

One practical AI idea, every week.

Real ways I use AI at work and at home. No hype, no jargon, no tech background needed. Coming soon for subscribers: the free course, 5 Days to Actually Using AI.

Free forever. Unsubscribe anytime.