What Is Marketing Science

Data Neighbor Podcast

0:00

-59:30

What Is Marketing Science

And How It Stops Wasted Spend

, and

Apr 11, 2025

Every marketer has heard the old adage: “Half of my marketing spend is wasted—I just don’t know which half.”

In this Data Neighbor Podcast episode, we finally get an answer from someone who does know. We sat down with Agastya Komarraju, Global Head of Product and Science for Growth and Funnel, ITA at Amazon, to unpack what “marketing science” actually is - and its role in shaping how companies decide what ads to run, where to spend, and who to target.

This isn't just about dashboards and metrics. It’s about how data, experimentation, and machine learning are radically transforming marketing into a measurable, testable, and ROI-driven function. And in Agastya’s case, how it led to a 90% cut in marketing spend for hiring - without sacrificing results.

What You'll Learn

The 5 Pillars of Marketing Science (and why every data or marketing team needs to know them)
Why traditional propensity models often fail, and how “persuasion modeling” gives you a more realistic view of customer behavior
Why offline marketing channels like direct mail still work, and how to tell when they’re actually worth it
The difference between Multi-Touch Attribution (MTA) and Media Mix Modeling (MMM) - and when to use each
How GenAI is shifting the landscape of creative development and copywriting
How to convince stakeholders when the model says one thing—but your budget owner says another

Key Takeaways

Marketing science = marketing with a PhD.
It’s not just campaign tracking. It’s experimentation, causal inference, and optimization all rolled into a function that connects data to business impact.
Don’t fall for the “sure shots.”
Sending offers to high-likelihood customers often wastes money. Focus on the persuadables - those who wouldn't convert without the nudge.
Attribution is messy, but necessary.
Whether you're running first-click, last-touch, or network DAG-based models, attribution only works when you validate it through live testing.
Offline channels can surprise you.
Direct mail isn’t dead - it’s just been poorly targeted. When used smartly (with the right modeling), it can outperform digital.
Real impact takes centralized strategy.
Agastya’s team at Amazon transitioned from a decentralized job marketing approach to a centralized, model-driven one - cutting waste while hitting hiring goals.
The future is GenAI, but marketing science isn’t going away.
While GenAI is transforming creative workflows, the real power lies in combining AI with science-led frameworks and human judgment.
Storytelling is a superpower.
Data is great. Insights are better. But the ability to communicate your findings in a way that moves teams forward? That’s the difference-maker.