The concept of ‘FAIL FAST’ comes up in client conversations from time to time. I’ve often wanted to distance myself from the concept in favour of ‘SAFE-TO-FAIL’ – a concept I’ve learned about from Dave Snowden. Often, ‘fail fast’ seems like a platitude used in businesses. The right thing for people to say is to get general agreement around the table and then do nothing about it. However, rather than trying to contrast the two concepts in an ‘either/or’ manner, I’m looking at how they could work together.
The paradigm with ‘failing fast’ is to have as rapid a feedback loop as possible. Not wasting time and resources on something that isn’t going to work. Sounds logical. In practice, what stops people from embracing ‘fail fast’ is a fear of the consequence of failure. It doesn’t feel safe, especially in the face of uncertainty and unpredictability. This is where ‘safe-to-fail’ comes in. It deals with how we design our actions with an emphasis on safety. Small or cheap enough that we can recover from failure without major consequence. Doing as little as possible to maximise our learning about an opportunity or risk.
We want to avoid an undisciplined or unmonitored approach to “building a culture of fast failure” or “celebrating failure.” Culture develops through many interactions over time. The practices that shape the interactions we want are:
- pre-mortem – careful consideration of what potential signs of failure could be,
- the sizing of our action based on our ability to monitor its impact and
- processes to exit or move away from the action if we see signs it isn’t working.
- This is how we minimise risk in the pursuit of learning. And, indeed, how we build on what we see working.
The focus for ‘fail fast’ is the speed of feedback, rapid testing of a hypothesis. Safe-to-fail minimises the risks of unintended consequences and can encompass the ‘fail fast’ mode. Yet, it is also used for exploration of hunches. Trends or signals in the market we haven’t formed a hypothesis about yet. Learning through action rather than analysis.
Here’s how it can work using a retail context. The process can apply in most sectors. We have a hypothesis that a new store layout could increase purchases. Or offering a new line of product will grow our market share. We decide how much time and resources we need to put in to give us enough data to know whether it is or could be successful. Yet, the investment needs to be an acceptable loss of time and resources (and reputation) if it doesn’t work. Figuring out the granularity – or right size – of the experiment. We could introduce the new layout or product line in 3 stores rather than, say, across all 20. Definitely more than one experiment at a time to learn from parallel experiences. Success can be both hard metrics and observable customer behaviour change.
We make sure we have the ability to monitor the impact of the initiatives we run. Have the ability to capture the data we’re looking for and the people in place to observe what happens. The leaders or teams meet often to make sense, as a group, of what they are seeing. How often we meet is a judgement call based on the experiment. If the initiative (or aspects of it) shows signs of success we amplify that success. We might widen the initiative to a range of stores. If we can see it is not working we start the off-ramp or exit actions. For example, we remove the products from the shelves immediately if we see a strong negative response from customers.
This way we can fail fast and minimise risk at the same time while maximising learning. Most importantly they can take action on uncertain ideas without having to make the action fail-safe i.e. too risky to fail. Due to the attention paid to monitoring, we will pick up on signals of risk and opportunity we hadn’t considered beforehand. In this way it’s not simply failing fast on a known hypothesis, it’s opening up a much broader spectrum of learning about our operating situation.
Enabling a safe-to-fail approach builds a culture where people can take action in the face of uncertainty. We avoid false assumptions or predictions that the action will work. We figure out what can work as we go – which can be fast.
This is a thought piece to explore my own thinking in response to a client situation I was dealing with. At AGLX we work with our clients on designing actions that are safe-to-fail and developing the practices that allow a culture of learning through action to emerge.
Get in touch if you would like to have a conversation to extend your learning.