A few years ago, I had the chance to work with a leader who had previously been at Amazon, and had brought with him a good chunk of the company’s culture: blunt feedback, big ambitions, and a relentless search for excellence.
Working with him was not always easy — but in a good way. He could be demanding, sometimes uncomfortable to deal with, but he had a way of raising the bar for everyone around him.
Although we weren’t even in the same reporting line — he was a Product exec, I was in Engineering management — we ended up having semi-regular walk-and-talk conversations. Strolling around Paris, we’d unblock operational issues as much as get to know each other. Sometimes, he would provide the best kind of unsolicited feedback: the one that makes you grow, and sticks with you years later.
One day, I was walking him through an initiative I wanted his opinion on. I was going deep into the details: trade-offs, risks, team concerns, possible consequences to the roadmap.
At some point, he stopped me and asked:
“Pablo, is this a one-way door or a two-way door decision?”
I stared back blankly at him. I had never heard the concept before.
He explained that one of Amazon’s cultural principles is bias for action, meaning that they value calculated risk-taking. For that, they use a mental model called one-way and two-way doors.
A one-way door decision is difficult to reverse. Once you go through it, going back is expensive, painful, or sometimes outright impossible. Large architectural migrations, shutting down a product line, committing to a long-term platform strategy, or deeply restructuring teams usually fall into that category. Those decisions deserve careful analysis.
A two-way door decision is different. You can try something, observe the outcome, and reverse course if needed. Most experiments, process tweaks, tooling choices, feature flags, or small refactors are two-way doors. In fact, most decisions we make actually are.
After he finished explaining, he looked at me and said:
“If this is a two-way door decision, then I trust you’ll do the right thing.”
He wasn’t dismissing the problem. He was teaching me where leadership attention should go.
Faced with a reversible decision where there is enough evidence to think it will benefit customers, it is often more expensive to overthink it than to just go for it, even if it turns out to be a bad idea in the end. Innovation is a trial-and-error business, and finishing second is deadly. Every minute spent overanalyzing a two-way door is time not spent on the one-way doors that actually deserve deep thinking.
As engineers, we have a bias towards precision: we want all our decisions to be the best decision we could have made. That instinct is useful, but searching for complete certainty can slow down innovation to the point that it ends up costing more than any mistake would have.
Part of leadership is recognizing where rigor creates value, and where it merely creates friction.
Spend time refining the decisions that are difficult to reverse.
For the rest, create guardrails, trust your team, and just do it.