MACHINE LEARNING FOR DECISION MAKERS
CHAPTER

5

Building a Data Science Workflow

And why businesses should look for a simple and automated integration.

David Mora
|
6 MIN READING TIME

When using ML for business decisions, it’s not enough to have the technology. You need to expose it to your team in a way that fits how they already work and think.

A thriving machine learning culture has automated updates, fast iteration, and outcome simulation

  1. Automated prediction delivery and ingestion of new data keeps data fresh and relevant. This frees up your team’s time to focus on innovation, not tedious data updates.
  2. When trying out new machine learning models is easy, predictions quickly improve, adapt, and harness the insight of varied stakeholders. It also removes costly lag time between data practitioners and the rest of your team.
  3. Machine learning models learn subtle relationships in your data. This allows you to predict individual outcomes. But how might big-picture changes affect the full landscape of your key outcomes? Innovative techniques now enable you to simulate the outcomes of major business decisions. This approach is bleeding-edge, and few public sources even discuss it. However, for companies who’ve adopted it, it’s become their competitive advantage and the closest thing they’ve had to a crystal ball.


A day in the life of a thriving machine learning culture

It's Monday morning. You open the email you open first every day: "New sales lead qualification predictions."

The attached spreadsheet shows new leads along with a machine learning prediction of how likely they are to convert to customers. After a quick column sort, you now know exactly what you should focus on for the day.

You have an idea for a new piece of data that might help make predictions even more accurate. You're not a data scientist, but within five minutes, you’ve uploaded the new data and created a new machine learning model. It’s a strong improvement, and you did it using excel!

At a meeting, you use your improved machine learning model to simulate how changes in marketing strategies can shape sales lead quality. When you tweak a parameter, like choice of marketing platform, the machine learning model predicts its likely impact on sales. Your team gains new understandings of where to focus their efforts, and the levers that will drive change.


Start quickly with a powerful and user-friendly ML tool rather than building from scratch.

By starting with a powerful tool that doesn’t necessitate code or a data scientists:

  • You’ll quickly learn if and how machine learning helps your business, and what data you need.
  • While still in the exploratory phase, the ability to quickly build and test models will save time and money. Relying on code or data scientists, particularly when starting, will severely slow your learning.
  • The user-friendliness and automation of the tool will set a clear benchmark for future data work and cross-team engagement with data.
NEXT CHAPTER
THE END