How startups should measure Marketing

👋 Hey, it’s Sundar! Welcome to a  paid subscriber only article of experiMENTAL: a weekly newsletter on B2C Marketing & data science how-to guides, frameworks, and stories from 15 years including early Uber. Free tier subscribers can continue reading for limited preview of the article.This week is sponsored by Superhuman AI

Why startups complexify measurement

Every startup wonders the same thing: "How do I know if my Marketing is working?" Seems like such a simple question, but the answer is not. If it was, then Martech wouldn’t be a multi billion industry.

Worse, the marketing measurement industry has made billions selling you the lie that you need fancy attribution models, complex dashboards, and a team of data scientists to understand if your Marketing works. Nope. I’ve now worked at one of the best marketing science teams in the world and then 4 startups in a row. Surprisingly, it’s startups that think marketing measurement needs to be more complicated than it needs to be and it happens for 3 reasons.

Illusions of:

1. Complexity

2. Certainty

3. Control

The illusion of complexity

When it comes to measurement, startups try to copy what big companies do. I understand the impulse. You want to be like Google, Meta, and Uber, but it's like a 10-year-old kid trying to bench press 300 pounds because they saw The Rock do it.

What’s even crazier is that startups try to copy what big companies do without understanding what they do. So, let me pull back the curtain a bit.

Yes, large companies have complex systems and data scientists churning out models and all that, but at the simplest level, it’s how much did we spend and what did we get from it. That’s what every MBR and QBR is about. Imagine being in a room with 20 senior leaders and going into the fact that you’re using a bayesian time series with something something covariate adjustments blah blah colinearity . Doesn’t happen. All of that magic is behind the scenes.

At these larger companies, complex decisions are still being made with blended metrics because we understand the nuances behind these blended metrics. At Uber, we knew that paid represented ~X% of our Blended CAC. We did that with incrementality testing and MMMs and other tools but we didn’t sit there and calculate a paid CAC and then our organic contribution…. We used a BLENDED CAC.

The illusion of control

Become a paid subscriber to continue reading.