The first A/B test ever

The first known A/B test lasted ten days. The prize for winning was more vegetables. The prize for losing was death.

The story opens with a victorious king selecting four captives to serve in his court. He told the guard to feed them a royal diet of meat and wine so they would remain strong.

The leader of the four declined. He asked for water and vegetables.

The guard didn’t care about the menu, but if these men wasted away malnourished, the king would have his head.

So the prisoner made an offer: “For ten days give us water and veg, and let the other prisoners eat the king's food. Then we’ll see who looks healthier.”

Test group, control group, dependent variable.

The guard took the bet. (Brave chap!)

Ten days later, everyone agreed the men on water and veg looked healthier than the other prisoners.

This was not a proper A/B test. No randomization, small sample, loads of confounding variables, and "looks healthier" is rather subjective.

To be fair, it happened 2,600 years ago in Babylon. (Daniel 1:6 - 1:16.)

Why am I telling you this? Because experiments are not new.

Yet here we are in 2026, making huge decisions based on theories and assumptions we could easily test with low stakes.

Why don’t leaders test their big ideas?

Some people blame hubris, but I reckon it’s fear. We’ve staked our reputations on our ideas, products, and pitches. Testing an idea means admitting it might be wrong.

But here’s the thing… experiments don’t cause mistakes. They reveal them. Our assumptions are already flawed, whether we test them or not. The only question is whether we find out in a small experiment or a big public failure.

As for the guard, he wasn’t afraid of being wrong. He was afraid his boss would kill him. If you can't risk being wrong in front of your team, do you really expect them to have the courage to be wrong in front of you?

Simple next step

What’s one high-stakes project you’re doing right now that you could test with a low stakes experiment? Here's my PDF with 20 scrappy growth experiments you can run with almost no audience, budget, or tech. Validate ideas before you build them. No vegetables required.