Building a data-driven culture

Demystifying Decision Science

Howard Friedman

Adjunct Professor at Columbia University

Unmet potential

Having data isn't enough - you need the right culture

Many projects get stuck, go unused, or lose stakeholder trust
Recognizing these early helps you avoid them
Common pitfalls fall into two groups: data-related and business-related

Lack of usable or accessible data
Insufficient infrastructure or tooling
Not enough skilled personnel
Poor quality control processes

No executive buy-in or sponsorship
Weak project management or unclear ownership
Insufficient communication
Inadequate investment
Poor collaboration between teams
No established decision science culture

Five key steps

Learn the basic vocabulary
Identify meaningful business problems
Ask good questions
Challenge assumptions
Embrace a test-and-learn mindset

One, two, three

Step 1: Master the language

Understand key concepts like risk, bias, and experimental design
Vocabulary builds shared understanding between business and technical teams

Step 2: Prioritize the right problems

Start with the business challenge, not the tool
Focus on problems that are both critical and solvable

Step 3: Ask powerful questions

Explore the goals, data, models, and assumptions
Push beyond the surface to uncover what truly matters

Four and five

Step 4: Challenge assumptions

Question the status quo and uncover hidden biases
Use data to test assumptions, spark innovation, and improve decisions

Step 5: Embrace a test-and-learn mindset

Not every experiment will succeed - and that's okay
Focus on iteration, learning, and long-term growth

More than models: It's about empowering everyone to make better choices

Practice makes perfect!

Let's practice!

Demystifying Decision Science

Preparing Video For Download...