Lessons from AWS: One-way and Two-way Doors

Lessons from AWS: One-way and Two-way Doors

Amazon, like possibly all big companies, is full of wisdom, and I was able to experience this firsthand during my time there.

I was not able to appreciate it back then, but every passing month I start to apply the knowledge and the lessons I learned there.

The one-way and two-way door lesson is one of choices and actions. You can read the full letter from Jeff Bezos to Amazon shareholders that started this here.

One-way doors are those actions that are irreversible, or very difficult to reverse. Examples include selling your company, or big rebranding operations where you change the name of the company. One-way doors often either carry a lot of momentum, or may require that, once the decision is taken, the company assumes it and moves forward with the choice.

Two-way doors are actions that are easy to reverse. These actions can sometimes be seen everywhere throughout the standard software development lifecycle. From A/B tests, blue/green deployments, experiments (like what Booking.com does) and so on. These decisions can be quickly reversed, and in the case of experiments, they’re deployed to a small fraction of their customer base to measure their impact. This is a great way to ensure you are building a great product iteratively.

The theory is easy enough, but actually identifying one-way and two-way doors in real-life situations is a much more difficult task. Often times, these decisions will not be as clear-cut as the examples given above. However, this will come to you naturally with practice. Being able to differentiate between one-way and two-way doors will make you a better thinker, philosopher, and leader. And if you are a software engineer, you will be able to take decisions in a much more agile way, as you will be able to quickly measure and understand the impact of your actions.

When it refers to businesses, banks used to be very slow-moving, and hesitant to adopt new technologies and systems, relying mostly on one-way doors, whereas startups used to be hectic and chaotic, treating everything as a two-way door, but it seems that with time, regulations have forced banks to adopt more modern technologies, like what PSD2 outlines, for security reasons, whereas startups are starting to build a resiliency culture possibly borrowed from banks, for their own survival.

Naturally, all of this is highly subjective, and often depends on the risk tolerance of the individual taking the decision. Sometimes, losing a customer target demographic is not a one-way door if it can be regained back, for example, when rolling out controversial marketing campaigns.

With that being said, it is also sometimes possible to make what would be one-way doors suddenly become two-way. Precisely, experiments and other highly specific business intelligence can inform your decisions, and either adapt changes, or inform you on what your customer base is more tolerant of, thus making it become two-way, as certain demographics are more accepting of change than others.

I encourage you, dear reader, to think about whether the decision you are about to take constitutes a one-way or a two-way door. If it turns out to be a two-way door, you will be able to take the decision without much decision fatigue. Simply commit to it, and undo it if you don’t get the expected results. There is no need to overanalyse every decision, which frees up valuable mental resources for the one-way doors that you will find in the future.