The Assumptions
Building Right by Proving Ourselves Wrong
After a little hiatus from Substack, I’m back to logging my daily thoughts.
We’ve finally settled into our new place in San Francisco! Now that we’re diving back into work, one of our main goals is to build a solid framework for testing assumptions, which is what I want to share today.
When we started dotflo, we didn’t think much about assumptions—let alone test them rigorously. We basically went straight to what we thought was the “truth” and built from there. In retrospect, testing assumptions early on would’ve saved us months. For example, when we worked on helping nonprofits find the right targets, we assumed they were already spending thousands on dedicated targeting teams. A few quick calls and right questions asked would’ve told us that wasn’t true. Or the time we built a research tool for SDRs, assuming they spent hours on research. Turned out, only the top-performing SDRs did, while the rest focused purely on numbers (as they were compensated on this). We spent months building for the wrong audience.
Lesson learned: test assumptions early. The cost of not doing so? Wasted time and energy. A couple of notes from the big one out there - companies like Airbnb and Uber — they started by testing simple assumptions. Airbnb’s: Strangers will pay to stay in someone’s home. Uber’s: People will book rides from private drivers through an app. Both tests began small. Airbnb initially tested this by renting out air mattresses in their own apartment during a conference in San Francisco and then a very simple website and finally the software. Uber just connected existing drivers with riders and tested their assumption in San Francisco.
Another thing we’ve realized is that assumptions often sneak in disguised as “facts.” It’s easy to fall into the trap of thinking that just because something seems obvious to us, it’s universally true. But users don’t think the way we do, and that’s the whole point. Assumption testing helps us see the world from their eyes.
We also learned that it’s better to test early and imperfectly than not at all. You don’t need an elaborate setup. The more straightforward your test, the faster you get real feedback. This was something we struggled with at first—thinking we needed to perfect the product before showing it to anyone. But the sooner you let go of perfectionism, the sooner you can course-correct.
As we head into the week, assumption testing is going to be front and center. One more thing: I’ve noticed a shift in how I handle feedback.
Before: I just wanted users to love everything we built and loved product validation.
Now: I’m actively looking to prove our assumption wrong. It’s all about asking the right questions.
~11 PM, Market Street, San Francisco (Under Moonlight)
