What Is Variation Scouting?
Variation shows where to improve. Averages make it harder to see. Here's why I started this series and what I'm trying to find out.
Averages Hide the Truth
When someone tells you “average delivery time is 3 days,” that sounds fine. But what if 40% arrive in 1 day and 60% arrive in 4.3 days? The average is the same, but the customer experience is completely different.
Variation is where the real story lives.
Why AI Tools?
Quality professionals have had great statistical software for decades — Minitab, JMP, R. But now AI tools like Claude, Copilot, and Gemini promise to make data analysis accessible to everyone.
The question is: can you trust them?
That’s what Variation Scouting is about. I take the same dataset, run it through different AI tools, and compare the results against proper statistical analysis. Same data, same questions, different tools.
The Dataset
I use Finnish railway punctuality data from Joensuu station — it’s real, publicly available (Fintraffic/Digitraffic, CC BY 4.0), and has enough variation to make things interesting.
What I’m Looking For
In each episode, I evaluate:
- Accuracy — Does the tool get the numbers right?
- Insight — Does it find meaningful patterns?
- Trustworthiness — Would you rely on this analysis for a decision?
- Usability — How easy was it to get useful results?
Follow along on YouTube to get weekly variation insights.