553 Startups, One Randomized Trial: Trained Founders Quit Sooner. The Ones Who Stayed Raised More Money.

Here's the brutal truth. Most of us don't fail fast. We fail slow, and we pay rent on it for years.

A working paper released through NBER in January 2026 tested this with a real experiment. Researchers from Rice, Georgia Tech, UC San Diego, the University of Houston and the University of Delaware randomly assigned 553 science and tech startups in 12 US co-working spaces. Some founders got intensive, short-term entrepreneurship training. Others didn't.

The trained startups were more likely to shut down, and they shut down sooner.

That sounds like bad news until you look at who stayed. Among the survivors, trained startups were more likely to raise outside funding, raised it faster, and raised more. They also reported higher employment and revenue. The authors say the results fit the idea that good training helps founders with weaker ventures quit rationally.

A few honest caveats. It's a working paper, not peer-reviewed yet. These were tech startups, not brokerages or agent teams. And the subgroup analysis is labeled exploratory.

But think about it like this. Every business carries a dead project. The farm area that hasn't produced a closing in 18 months. The paid lead source you keep because you already paid for it. The team member you're still hoping will turn the corner.

The founders in this study didn't get braver. They got clearer, and clarity made quitting the cheap option.

So here's your assignment. Write down the one thing in your business you'd stop today if you were starting from scratch. Give it a kill date and a single number it has to hit by then. Miss the number, shut it down, and move that time to what's already working.

Originally published by Esther Bailey, Daniel Fehder, Eric Floyd, Yael Hochberg and Daniel J. Lee as NBER Working Paper 34755. Read the full paper here: https://www.nber.org/papers/w34755