Tony Xu with David Senra
Inspired by Tony Xu on David Senra
Worth Stealing
The earned secret
Sometimes, the most valuable insights cannot be modelled in advance. DoorDash discovered suburbs beat cities for delivery density only by trying both. Go out, experiment and earn the insights.
Invisible data is the real risk
The data you can see, you act on. The data you cannot see is what kills the business. Know what information you have and actively design ways to collect what you do not.
The tail of the distribution is the product roadmap
Anecdotes clash with averages because they sit at the edges. But the edges are where the next product ideas live. DoorDash moved from food delivery into helping restaurants acquire customers and manage stock by listening to the outliers, not the median.
43 minutes to a real test
The first version was a static page, eight PDFs and a Google Voice number. The point was not to build something cheap. It was to get to a real answer as fast as possible. Does anyone actually want this?
Experimentation is the operating system
Xu does not treat experiments as a phase. They are how the company thinks. You do things that do not scale to collect ideas, then build products that do.
Hire for action, not analysis
The final round interview: $20, eight hours, go acquire 100 customers. A test of whether someone would move, not whether they could model.
Just get started
Know your next two or three steps. Not the whole plan. The rest reveals itself through doing.
My Thoughts
Forty three minutes. That is how long DoorDash's first version took to build. A static page, eight PDFs of restaurant menus, a Google Voice number that rang a founder's phone. Someone would call, read their order, and a founder would drive it over. No app, no logistics layer, no rating system.
It is probably one of the most extreme MVP ever shipped. And the point was not the product. It was the question: does anyone actually want this?
The tools have changed. With Cursor or Lovable, you can put something genuinely polished in front of a customer in an afternoon. The quality floor of an MVP has risen sharply. But the underlying logic has not moved at all. You are still trying to answer the same question as fast as possible, with the least amount of built assumptions baked in.
There is something clarifying about the DoorDash version precisely because it was so bare. If someone called that Google Voice number, placed an order, and came back the next day, you knew.
No interface, no convenience features, no network of drivers to smooth things over. Just: do they want the thing?
That is the test worth running. If a customer loves the product before you have added the bells and whistles, they will love it more once you do. If they only love it after, you have built yourself a dependency and you do not yet know whether the core idea holds.
Most founders delay shipping because the product is not ready. DoorDash shipped before the product existed. The gap between those two positions is where most founders lose months they do not get back.