Hypothesis-Driven Discovery
by Jag Duggal • Chief Product Officer (CPO) at Nubank
Jag Duggal is the Chief Product Officer at Nubank, leading product strategy for one of the world's largest digital banking platforms. Previously, he served as Director of Product Management at Facebook (Meta) and held senior strategy and product roles at Quantcast and Google.
🎙️ Episode Context
Jag Duggal discusses how Nubank grew to be larger than major US fintechs combined through word-of-mouth and a fanatical focus on customer love. He details their strict criteria for scaling products based on modified Sean Ellis scores, the importance of being "fundamentally different" rather than incrementally better, and how to execute hypothesis-driven customer research.
Problem It Solves
Prevents wasted research time and ambiguous data that leads to no clear product direction.
Framework Overview
Effective discovery starts with a crisp hypothesis. Researchers act as judges (impartially testing the idea) rather than lawyers (trying to prove the idea works). Direct observation and small sample sizes (calling 10 people) often beat large datasets.
🧠 Framework Structure
Formulate a crisp hypothesis before t...
Act as a judge of the hypothesis, not...
Observe behavior rather than just ask...
Pick up the phone: PMs should persona...
When to Use
During the early stages of product definition or when trying to understand why a launched product isn't performing.
Common Mistakes
Falling in love with the hypothesis and conducting research just to validate confirmation bias.
Real World Example
Nubank hypothesized that joint accounts are 150-year-old artifacts based on outdated laws. They are researching modern 'social finance' (sharing goals/bills) rather than just digitizing the old joint account model.
If you don't have a hypothesis, you're going to spend a lot of time researching... and you're not going to know what to make of it.
— Jag Duggal