It used to be easier when I was working in the factory. If we wanted to make a change to a production process we could usually determine fairly quickly whether the change led to a real improvement in performance. We had control over all the inputs and process variables. We were building over 2 million units a month, so the before-and-after samples could be large enough to perform a rigorous hypothesis test. The tough part was convincing people to wait patiently while the “after” sample was measured and analyzed.
Business process improvement is more challenging. It’s harder to control the inputs and variables, or understand all the external influences, which means it’s harder to model the process and figure out how to improve it without breaking something else. It’s harder to run experiments to test proposed improvements. The improvements usually require training and software changes that require some time to implement. It can take weeks, months, or even years to really determine whether the changes are effective. During that time you may get persistent reports of process problems that pre-date your improvements (“I thought you were fixing that process!”).
Process improvement requires time and patience, and stakeholders and sponsors may not be able to afford either. In 1933 during the depths of the Great Depression, with unemployment at record levels, US President Franklin Roosevelt and the US Congress implemented a haphazard and sometimes contradictory series of programs (the New Deal) to stabilize and improve the economy. This wasn’t the time for study and controlled experiments. Some programs worked better than others, and it’s been suggested that the New Deal wasn’t even all that effective in ending the Depression. Nevertheless, this was an extreme “launch and learn” situation.
The failure of a business process may not be that kind of crisis, but it’s important to set realistic expectations before embarking on an improvement effort. When is improvement required? How much improvement is needed? What level of performance is “good enough?” How will it be measured? What’s the balance between achieving quick results with a temporary solution vs. implementing an analytically proven solution?
The answers to these questions will help you understand how much time you have, what success looks like, and how much patience you can expect.
