So his conclusion is that you need to get good at trying new things / shoot goals, because it DOES come down to "I have no idea what will stick". But the more shots you shoot, the more likely.
But also— most of us are in software, and every product we build (including Segment) has compounding effects towards the next thing we build. The frameworks we use or build can inform our next thing. The account flows. The databases and CI/CD we get used to. For every thing we try, we get a little better at the next thing.
I've found that "launching ideas" from a napkin sketch to working prototype has become significantly easier. Used to take me a month+ to launch a crud idea/prototype; now it takes me a weekend.
So the limiting factor for me isn't building/launching anymore... it's customer discovery and sales.