Stop Being a Data Scientist. Start Being a Data Craftsman.

Businesses are drowning in data but starving for insights that transform their bottom line. Why? Because we treat data science like a production line, not a craft.

Data Science
Machine Learning
AI
Business Strategy
Author
Affiliation
Published

3 August 2025

Businesses are drowning in data but starving for insights that transform their bottom line. Why?

Because we treat data science like a production line, not a craft.

The Problem: Testing does not trump insights

Marketing was perhaps the first and the worst offender. Rigour in analysis and understanding were downplayed in favour of A/B testing, often with vanity metrics. I get it. Except the vanity metrics.

I may not have invented A/B testing but I did create1 an approach to marketing called Customer Value Management (CVM) that was heavy on the automation while closely monitoring and optimising the short and long term financials. (The key innovations were the change process, the links to financials, and the processes for sustaining innovation.)

https://www.cybaea.net/CVM.html

It was great, for its time. We went from not using data to using data; from decisions based on hunches and whoever was paid the most to decisions based on data; from marketing vanity metrics to financial metrics that actually matter for the organization. We reliably increased profits by 5%, every year.

Too few companies today do even this.

Aside: Stop the data envy already

We were lucky to do this in mobile telecommunications. This was the industry that had loads of data on their customers and their behaviours. My definition of Big Data is that you have it when you are forced to throw away more data than you are able to keep. Mobile telcos with their detailed network information checked that box.

Other sectors were envious of our rich data, but that was the wrong obsession: They did nothing (very smart) with the data they had.

Having lots of data meant that we could try lots of things, but also that there is a strong temptation to ‘just do data’ and spend less energy on understanding.

Fast forward twenty years since that first CVM implementation at Vodafone Netherlands, and today every business has a lot more data than it used to possess.

And too many of them are still looking with envy at other businesses with more data, using that as an excuse for not doing enough with what they have.

Enough with the excuses already.

The insight: AI can do the testing work but not replace humans

Every business now has more data. And the “easy” data work—the simple A/B tests and vanity metrics that were once common in marketing—is being automated by #AI. Those jobs are mostly gone, and AI will replace the remaining ones.

So, how do you remain relevant?

And how do businesses gain true Return on Data? How do we deliver the impact that transforms our performance?

The Action: Rediscover craftsmanship

We remain relevant by embracing craftsmanship. By going beyond the algorithms to deeply understand the data, the problem, and the business.

If you are doing data science and you are not reading “Applied Machine Learning for Tabular Data”, then you are doing it wrong.

https://aml4td.org/, now available up to chapter 15.

It is a fantastic resource because it teaches you to go beyond and behind the statistics and algorithms to help you understand your data and thereby ultimately whatever the data is about. Say, your customers.

They treat machine learning as a holistic and practical craft, rather than a purely theoretical or algorithm-focused discipline. The emphasis is on the entire modelling pipeline, with a strong belief that the steps before and after model fitting are just as important, if not more so, than the choice of the algorithm itself.

This is not a ‘cookbook’ of oversimplified rules. It is not a ‘black box’ approach promoting the authors’ pet algorithm (or, worse, trying to sell you a proprietary one).

Instead this is for the pragmatic practitioner. It is a book about the craftsmanship of data science.

Aside: Learn from the best

You are doing it wrong, first because Max Kuhn is super smart2 and someone from whom you want and need to learn.

Second, because the level of understanding of the data, the problem, and the algorithms you get from following the methodologies you learn here are exactly what you need to remain relevant in a world where #AI promises or threatens to automate the easy work.

Max comes from pharma and clinical trials which always required deep understanding. Unlike, say, marketing.

Image from chapter 9: Overfitting of aml4dt

The Impact: Craft + curiosity finds opportunities and delivers change

Whether you read the book or not, please take this away:

  • Getting the right insights from data is a craft that requires curious humans.

  • Getting the right business outcomes requires humans to build connections, understand opportunities, and deploy solutions across the organization.

The Bottom Line: Curiosity didn’t kill the cat, it won the race

We include AI in all our products these days. Not to replace humans but to make us more curious and adventurous.

☛ How are you building your data craftsmanship to stay ahead of the AI curve? Share your thoughts below. ☚


About me: I help organisations turn complex data into clear decisions and commercial outcomes. My focus is on enabling better decision-making and unlocking new value through data-driven innovation – especially where the stakes are high and the problems are difficult and poorly defined.

Follow me on LinkedIn for more insights.

AI generated image of a Data Craftsman

Footnotes

  1. Primarily with James Wilkinson and Martin Dixon-Tyrer and strongly supported by Cretien Brandsma and others at Vodafone. See some case studies here: https://www.cybaea.net/Case-Studies/.↩︎

  2. I’ve never met Kjell Johnson. He is probably a great guy.↩︎