What a flopped training program taught me about innovation and growth
This article was first published in the EPIQ Success newsletter. I am cross-posting it here for your enjoyment.
A few years ago, my colleague Elías Díaz and I built an EQ training program that flopped.
We’d previously had success with EQ keynotes and live workshops inside Amazon, but we wanted to hyperscale and reach far more people than human facilitators could. So we designed Learning Circles: small cohorts, meeting weekly, working through an 8-week emotional intelligence curriculum together. It was social learning with regular touch points, held together by real relationships.
On paper, it was a smart design. You build skills like emotional management, empathy, and influence best in community, not alone watching videos.
In practice, completion rates told a different story. People signed up excited in week one. By week four, half the group had dropped off. Embarrassingly few people completed the whole program. Turns out a weekly meeting, even a well-designed one, is a heavy ask for people who are already maxed out on bandwidth.
So we pivoted. Same 8-week curriculum, but async and mostly self-paced, with two live touchpoints instead of eight. Completion jumped significantly. And Elías and I learned something valuable: what looks great in theory can work completely differently under real-world conditions.
That failure taught me something I want to unpack with you today: not all failure is created equal.
The three kinds of failure
Amy Edmondson, the Harvard professor famous for her research on psychological safety, has spent her career studying how organizations innovate. Failure is an essential part of the process. But not just any failure, mind you—the right kind. She breaks failure into three categories, and I believe every leader needs to know the difference to build a culture of innovation.
- Basic (or preventable) failure is getting a routine process wrong. Following a standard operating procedure incorrectly. Skipping a step you already knew. The goal is have zero (or close to zero) of these. There’s no learning to harvest from forgetting to follow a checklist you’ve run a hundred times.
2. Complex failure shows up in any large, interdependent system: a hospital, an aircraft carrier, a data center. These systems will fail sometimes, no matter how good you are, because there are too many moving parts for perfect performance. The goal is to catch the small failures before they balloon into a big one. For example, most hospital accidents come from a chain of small failures that happened to line up in exactly the wrong way. At Amazon, we handled these with something called a Correction of Error (COE). You write a document that dives deep into why something happened, ask “why” five times to get past the surface explanation, and fix the system, not the person. You don’t blame people. You determine what conditions let the mistake happen in the first place, and you close that gap.
3. Intelligent failure is the one most organizations get wrong, because most leaders treat it exactly like the other two. It’s a thoughtful (and possibly bold) experiment that teaches you something you couldn’t have known any other way. This is the good kind. You want as many of these as you can afford, run at the right scale. It’s the entire point of working anywhere near a frontier.
My Learning Circles were an intelligent failure. We didn’t know the format wouldn’t hold up against real calendars until we ran it. Now we knew, and we built something better because of it.
One-way vs. two-way doors
You can’t build a culture that runs toward intelligent failure without also building a way to think about risk. Amazon’s answer is a simple but powerful mental model. Before you make almost any decision, ask a meta-question first: is this a one-way door or a two-way door?
A one-way door is irreversible. Once you walk through it, you can’t walk back. These decisions deserve real deliberation, a written narrative (aka six-pager) that forces rigorous thinking before you commit. A one-way door earns that caution for good reason: either it requires a huge amount of investment, or it has a large blast radius if it goes wrong (e.g., negative impact on customer trust, reputation, or revenue).
A two-way door is reversible. You can walk back through it if you’re wrong. For these, the instruction is the opposite: move fast. There’s a saying at Amazon that captures it well: if it’s a two-way door and you have 70% of the information that makes you think it will work—you’ve already waited too long.
This mental model is what lets an organization make both high-quality decisions (for one-way doors) and high-velocity decisions (for two-way doors). It lowers the cost of failure, which makes a high volume of intelligent failures possible.
Here is a powerful practice for leaders driving innovation: when a decision feels like a one-way door, ask whether you can turn it (or parts of it) into a two-way door. Can you shrink the blast radius? Reduce the complexity? Run a smaller version first? Often, what looks irreversible at first glance has a reversible version hiding within it.
How to build a culture of experimentation
None of this works as an inspirational poster on the wall. It only works if leaders lower the cost of failure and make experimentation expected, supported, and rewarded. There is no innovation without a culture of experimentation.
A few things you can start doing right away:
- Signal clearly, in words and behavior, that an intelligent failure is not a career-limiting move.
- Share your own failures. Your team is watching how you talk about yours before they’ll trust you with theirs.
- Celebrate intelligent failures out loud, the same way you’d celebrate a win, because the lesson is the win.
- Ask your people a simple question in your next 1:1: “What’s the best experiment you ran this week (or month)?” Not the best result. The best experiment.
In the end, not all failures are created equal. The leaders who figure out which kind they’re looking at—and build a culture that treats each one appropriately—are the ones who maximize their intelligent failures, and eventually, their biggest wins.
Be EPIQ,
Rich