receives compensation from some of the companies listed on this page. Advertising Disclosure


Five Steps to Conducting Safe-to-Fail Experiments in Business

Melanie Parish
Melanie Parish

During times of uncertainly, it's a good time for leaders to work toward change by conducting experiments that, while they may not work in the end, will not harm the company's bottom line.

As businesses are opening again, leaders need to think about what they are doing while so many things remain uncertain.  While the market mandates trying new things right now, how you experiment can determine your long term ability to succeed and thrive in the marketplace. Being an experimental leader means that you have a strategy in place to mitigate risk, to evaluate risk and to make decisions that put the risk at the right level as you experiment. It is an interesting time to contemplate what it means to conduct safe-to-fail experiments.

In order for an experiment to be effective, it must change something. If it changes too many things it has the possibility of putting your business at risk in some unexpected way. Safe-to-fail means the experiment will allow the leader to collect data without harming the business it is safe to fail. This means that if the experiment goes terribly, the business will remain unharmed. It allows the business to collect data and to create the next experiment so that it can move forward.

Here are five questions to ask yourself if you are wondering if your experiment is safe to fail.

What is the client impact?

If you're wondering if your experiment is safe to fail, you want to first consider the impact on your clients. And remember a term safe to fail. If your experiment absolutely cataclysmically fails, then what will the impact be on your client? This is particularly important if you're collecting data on customer uptake, or customer experience, or how your client feels about your product. This is particularly true if you are experimenting in a way that could have your client feel negatively toward your product. Changing the way your client feels about you may not be in the realm of safe to fail.

Example: Increasing frequency of updates of a software product from weekly to monthly. Do you have clients who have customizations that will require more frequent updates? Have you included a solution for them in your rollout?

How will this impact the business financially?

Some experiments have a cost and others do not. It is important to consider the financial ramifications of any experiment you do. It is important to take into account the worst-case scenario as that is what could happen if you fail. One way to mitigate cost is to reduce outlay of capital by using prototypes and other experiments to test the market to, to get proof of concept, and to try before you spend. On the other hand, sometimes organizations constrain themselves too much and avoid experimenting altogether and never invest in their own futures.

This experiment is destined for certain failure – a company that isn't trying new experiments will be stuck in the past. Experiments don't have to be cost-effective in order to be functional. We aren't always looking for efficiency first. Eliyahu Goldratt, an Israeli physicist that consulted in manufacturing and created "The Theory of Constraints" suggests you first optimize flow and then worry about efficiency. When you consider safe-to-fail experiments you want to be cost-effective and to have quick turnarounds on your data collection. This will help your experiments be safe to fail from a budget perspective.

Example: Using Facebook ads to bring in new business. Have you figured out a way to test your ads for conversion with a small spend before you make a big investment?

How will our staff feel?

As a long time business and executive coach, I don't believe how your staff feels about new initiatives should be your most important metric. But you do want to consider their reaction and their potential reaction to the experiments that you're planning before you roll them out. Will you get pushback? Will you get buy-in for the experiment? What will the impact be on your staff as you roll out the experiment? There are no right answers here. It is simply important to consider the impact of the experiments you do on this. Understanding the staff reactions before they happen can help you make your experiments so safe to fail before you roll it out.

Example: Rolling out a new dress code (every new rollout is an experiment). Have you considered how your staff will react? What will be their concerns? What is the business case for your dress code?

Are heroics involved?

When I think about "heroics" in a business context, I think about people going above and beyond to produce work. Are people giving more to their work than is expected and to the point where it might be on the edge of having a negative effect on them personally. We have all had projects we have gone all-in for in the short term. It is sometimes incredibly rewarding to ask a team to go all-in on a project. I believe that this extra effort can pay off in an experimental timeframe, but it needs to be used sparingly in the day-to-day operations of a business. Having one person work extra hard during an experiment needs to be considered as we ponder whether an experiment is safe to fail.

Extracting value from our staff and employees can have longterm consequences and may produce burnout. That burnout can have long-ranging repercussions on our business. We want to carefully consider whether our experiment requires heroics that take a personal toll on the people we work with. That is not to say that on occasion it isn't a very useful way of experimenting. We just want to make sure we are very careful about the expenditure of this kind of heroic effort with our people and on our teams. It could have a long-ranging impact on our organization, and we want to consider it carefully as we contemplate whether our experiment is safe to fail.

Example: Product launch. Have you planned how long you expect your staff to work long hours as part of the launch and when you expect them to return to normal?

Does it matter to anyone?

We have spent a lot of time talking about all the ways that we might need to be careful in the way that we experiment. It is also important to make sure that our experiment is big enough. Having a series of lukewarm and safe experiments may not allow us to move at the pace of change in the world as a whole. We need to be doing experiments that matter, that are relevant and that have the potential to produce impact in our businesses. Make sure your experiment is in an area that matters and has the potential to make a difference. Being too safe may be an experiment that isn't safe to fail. Being too safe may induce the failure we're trying to avoid.

Example: Rollout of software product features. Have you determined before you experiment how universal the need is for the new feature you are creating?

It is the job of the leader to figure out how to experiment in a way that keeps experiments safe to fail. The experiment shouldn't run for a long time – I always recommend 24 hours or seven days as a good starting period of time to help create the right size experiment that will foster a safe-to-fail environment.

It is the leader's job to ponder all the implications of an experiment on the rest of the business to make sure that the business is protected, that the experiment will have the desired impact, that it will collect the desired data, and that it will not produce unintended results. If your experiments are safe to fail, your experiments will foster a culture of innovation and forward movement.


Image Credit: fizkes / Getty Images
Melanie Parish
Melanie Parish Member
Master Certified Coach Melanie Parish is the author of the best-selling book, The Experimental Leader: Be a New Kind of Boss to Cultivate an Organization of Innovators. Melanie offers over two decades of working with Fortune 500, global IT and start ups. She is also a public speaker, popular podcast host, workshop leader, founder of Experimental Leader Academy and Prism Award winner. Her expertise is for any leader who has had to quickly pivot to become the new kind of boss this unprecedented time requires or really anyone who wants to step up and help lead.