Tempered Lean; Experimentation.

Lean Experiments are a way of testing ideas quickly and inexpensively. They also allow for testing an idea on a large scale - sometimes a simple process change makes for an amazing result. Experiments give us more information with which to make decisions. The key with Lean experiments is that they are relatively inexpensive, are designed to show results (positive or negative), and are easily reversible.

What is a "Lean Experiment"?

A lean experiment is just a test that we run to give us information about a certain concept or theory. We may have a hunch, based on experience, but we do not really know exactly what information we are going to get from the test until the end. We conduct the lean experiment in order to see if something works or not and whether it takes us towards throughput without any undesirable effects or UDE's. We don't want the experiment to blow up in our faces.

For example, if you wanted to test whether your SEO (Search Engine Optimization) strategies are effective, you might conduct a lean search experiment. A lean search experiment is an experiment that looks at how well your site appears for relevant searches in your target market.

The way you run a lean experiment is simple; create an experiment and carry out it. Some people may call this an "Event" but that makes it sound like something you do only once; with experiments, we run them every day. A lean experiment involves creating one, looking at and addressing any concerns or UDE's, deciding what data is captured, and carrying it out.

A lean experiment is generally a part of a larger test or project and it aligns with the goal (most companies need to make money now, as well as into the future). Your experiment might only last a few hours, a day, or it might last a few weeks. There are some typical questions around experiments and their answers below.

When should I start my experiment?

You should start your experiment once you have decided on a concept. Once you have decided, you should write up the learning objectives for the test. This will help to keep focused. What do we hope to achieve?

What should be in the learning objectives?

The learning objectives for your test should state what you are trying to achieve with the test. Are you trying to eliminate contamination? Are you looking at increasing throughput? Does a machine keep breaking and it's affecting the bottleneck?

When should I start my test?

You should start your test once you have decided on a concept AND you have written up the learning objectives for the test. This will help to keep focused.

I use a one page experiment form (Like Google forms). You could use a spreadsheet or any other way to document what you are doing, and why.

How will we implement the experiment?

We come up with a basic plan that accounts for all of the UDE's we have mitigated and we make sure it's communicated clearly. One of the unfortunate things in a process is running multiple experiments at once, whether it's intentional or not. This happens more than you think, and it contaminates your results in many processes.

What do we do at the end of the experiment?

We will run the experiment when we have collected the data and it shows a clear direction. Once we have the data, we will analyze the results. Often one experiment springboards into another. Pareto charts can also show where to place your efforts.

Whether it's a positive or a negative result, the experiment should be considered a success. Hey, at least now you know what NOT to do, right? And you can then devise a "plan B" to move in a different direction.

Eventually, through running many experiments, you will eventually steer your ship, your process, towards greater throughput with and changes in the wind effortlessly taking you towards your goal.