Dave Langer

📌 If you're ready to build the skills to unleash the power of Copilot in Excel, I want to help.

I have 11 hours of free crash courses:

Python
Logistic regression
K-means clustering
Decision tree machine learning

Join 8,587 preparing for the future: daveondata.com/newsletter

I won't bore you with the rest of the back-and-forth.

Here's the things Copilot assumed I knew:

K-means clustering
Silhouette score for optimizing k-means
Principal component analysis (PCA)
Decision tree ML models
Permutation importance
Box plots

You need DIY data science skills to use Copilot!

1 - Customer Segmentation
2 - Trend Analysis
3 - Predictive Modeling
4 - Basket Analysis
5 - Churn Analysis

When I asked for a recommendation, Copilot suggested option 1.

Copilot responded:

"...it appears to represent customer behavior metrics, possibly related to transactions, activities, or engagement levels."

I told Copilot to assume customer behaviors and asked it what kind of analysis it would suggest for the data.

Copilot responded with 5 options:

Copilot asked me to specify a deeper analysis.

I chose Copilot's correlation analysis suggestion.

In response, Copilot generated Python code to create a correlation heatmap table.

The correlation heatmap was interesting, but not super insightful.

I asked Copilot what the data might be.

To Copilot's credit, it replied that it didn't know what to do with the data.

Copilot offered some suggestions for various types of analyses:

Trends
Correlations
Predictive Modeling

As usual, it doesn't take long before Copilot assumes the user has DIY data science skills.

Next...

The table I used included data for:

Customer purchases
Customer demographics (e.g., age)

As usual, I chose Copilot's "Advanced analysis" option.

This option creates a new worksheet where Copilot generates Python code to perform the analyses.

Copilot loaded the table and profiled it...

Vibe Analytics with Copilot in Excel - Day 2

Copilot in Excel does an excellent job of using column names to understand the nature of your data.

So I decided to test Copilot in Excel using column names like:

Column1
Column2
Column3

Here's what happened...

I have 11 hours of free crash courses covering:

Python
Logistic regression
K-means cluster analysis
Decision tree machine learning

Join 8,857 professionals building skills to impress their leaders: daveondata.com/newsletter

Learning a subset of Python and data science handles 80+% of use cases.

Some examples:

K-means
Decision trees
Random forests

Here's the good news.

Any professional can learn these.

No advanced math required.

If you're ready to learn, I want to help you...

4) Recognize hallucinations.

All AI technologies make mistakes (hallucinations).

**BUT**

If you don't have the right skills, how can you vet Copilot's?

Customer segmentation
Fraud detection model
Sales forecasts

So, do you want to play with 🔥?

Or do you want to impress your leaders?
👇

3) Learn DIY data science.

You know that Copilot-generated code I just mentioned?

It's full of data science goodness:

K-means clustering
Decision tree machine learning
Time series forecasting models like ARIMA

This is how Copilot works its "magic."

There's a huge problem, though...

2) Learn Python.

Make no mistake.

What excites your leaders about Copilot in Excel is clicking a button to get AI-generated:

Data insights
Advanced analytics
Time series forecasts

You know what they all have in common?

Copilot-generated Python code.

But there's more.

1) Ignore "prompt engineering."

I've tested Copilot in Excel, and what it can do is impressive.

But here's the dirty little secret the AI bros won't tell you.

To prompt Copilot correctly, you have to know what you're doing.

You need to build your skills first.

Prompting comes after.

Want to differentiate yourself at work with Excel skills?

Leaders aren't excited about this in 2025:

- Using PivotTables
- Knowing many functions

You know what they're excited about:

AI

And that means Copilot in Excel.

Here are 4 ways to build skills to impress your leaders:

Want to predict customer churn?

Don't overcomplicate it.

Use logistic regression.

It's powerful, interpretable, and works great with real business data.

Start with Python in Excel.

📌 Ready to build skills for the AI-driven future of Microsoft Excel?

I have 11 hours of free crash courses:

Python
Logistic regression
K-means cluster analysis
Decision tree machine learning

Join 8,587 professionals building skills: daveondata.com/newsletter

Is your company rolling out Copilot in Excel?

Here's the mistake your executives are probably making:

To make the most out of AI in Microsoft Excel, you need to know 2 things:

1 - Python
2 - DIY data science

Copilot in Excel relies heavily on Python.
👇

📌 Ready to build skills for the future of Microsoft Excel?

I have 11 hours of free crash courses that will jumpstart your skills:

Python
Logistic regression
K-means cluster analysis
Decision tree machine learning

Join 8,587 professionals building skills this weekend: daveondata.com/newsletter

Excel → Power Query → Python → Machine Learning

That's the real progression for DIY data science.

You don't jump to neural networks.

You grow step by step.

From formulas to forecasts.

📌 P.S. - If you're ready to build skills for the future of Microsoft Excel, I want to help you.

I have 11 hours of free crash courses covering:

Python
Logistic regression
K-means cluster analysis
Decision tree machine learning

Get started this weekend: daveondata.com/newsletter

8,587 professionals will receive a new machine learning tutorial tomorrow:

ML with Python in Excel.

It will be the first in a series of tutorials showing how Microsoft Excel is the platform for DIY data science.

Want in?

Click the link in my profile to get started.

The Copilot AI inside Excel can write Python code for you.

That's the future Microsoft envisions.

Here's what Microsoft won't tell your executives.

If you can't prompt it well, you're playing with 🔥.

Learn enough Python to guide the AI...

And leave 99% of Excel users in your dust.

📌 If you're ready to build skills for the future of Excel, I want to help.

I've got 11 hours of free crash courses covering:

Python
Logistic regression
K-means cluster analysis
Decision tree machine learning

Join 8,587 professionals building skills: daveondata.com/newsletter

AI can make you 10x faster.

But you are still the expert in your business and your data.

Learn Python in Excel, even at a basic level.

That’s how you turn Copilot from a risk into your competitive advantage.

Python in Excel is your key.

Ready to take the wheel?
👇

If you just trust the output…

You might base a business decision on a broken forecast.

Or miss a looming downturn that Copilot’s model simply didn’t see.

Here’s the truth:

Copilot + Python in Excel is like giving you a race car.

But if you can’t steer, you’re just along for the ride.

A finance pro uses Copilot to “forecast next quarter’s revenue.”

Copilot generates a Python forecasting model.

Looks impressive.

But here’s what’s hidden:

- Key drivers like seasonality or product launches are missing

- The model assumes sales are always rising

- Outlier months are ignored

But here’s the danger nobody’s talking about:

If you don’t understand Python, you won’t realize what Copilot is really doing with your data.

Professionals who blindly use Copilot-generated Python code are playing with 🔥.

Consider this.

The real power of Copilot isn’t just its summaries.

It’s this:

Copilot writes Python code for you directly in Excel.

Suddenly, you can:

- Run advanced models
- Build custom visualizations
- Automate complex analysis

All with a simple prompt.

Microsoft has put Copilot AI right into Excel.

At first glance, it looks like magic:

“Summarize my data.”
“Build me a forecast.”
"Analyze these sales trends.”

But here’s what the hype misses: