Faster, risk-free and successful start-ups? A critical look at the Lean Startup
The development of the Lean Startup
In 2011, Eric Ries published his book "The Lean Startup" and coined the term "Lean Startup". The concept was promoted by the Redline publishing house as "founding companies quickly, without risk and successfully" and is now considered standard in lectures, among investors and accelerators. Nevertheless, many start-ups still fail, which indicates that the concept is being misapplied.
The Lean Startup is based on Steve Blank's Four Steps to Epiphany from 2007 and others, which form the basis of many dogmas accepted by startups today. Ries comes from a background of iterative processes, particularly design thinking. He adapts the first three steps into a cycle instead of a linear process and places the core of Lean Startups on the Build-Measure-Learn Cycle.
This says that we should quickly build a prototype, go to market with it to measure the impact. From this we learn for the next cycle, in which we build another prototype. In theory, this reduces the risk of going in the wrong direction and failing.
Before the concept of Lean Startup, long business plans were written before business validation, which contained all relevant information about customers, product and markets on several hundred pages. The obvious problem with this was that all this data was created in vitro. Together with other factors, this model ended up in the New Economy crisis. Some examples of companies that failed with this are Instacart and others.
Less risk thanks to agile development?
Studies show that today every startup uses the Lean Startup methodology, but this does not necessarily mean that startups have become less risky. To get to the bottom of de-risking, American data series were examined. The result shows that a Lean Startup is not less risky. In fact, the proportion of bankruptcies in innovation markets has remained constant since the introduction of Lean Startup. Similarly, the proportionate growth is not significantly different. This leads to the understanding that Lean Startup does not entail de-risking for the company as a whole.
One explanation why this theory does not work lies in the risk landscape. Let us imagine a hypothetical risk landscape with two decision dimensions and one risk dimension (z). To minimise risk, we make decisions at the decision level. A long business plan goes a long way here to get to places on the risk level that we cannot initially assess. Therefore, the value of business plans is indistinguishable from random decisions.
The idea of the Lean Startup is now incrementally moving into the step of the strongest gradient, i.e. the direction where we can reduce the risk the most. The study now shows that we are quickly moving into a local minimum, which is not optimal globally. To get out of this, we need a pivot that takes us to another place in the risk level. Since we again cannot know which minimums are desirable, we cannot achieve significant results on optimising risk with Lean Startup.
Nevertheless, the concept offers an advantage in that faster market feedback is possible and less value destruction is carried out because no long business plan is written with false assumptions. Risk management comes to the fore to minimise risk exposure. Ideally, there is either no return or a massive return. In other words, a successful startup creates a lot of value for all involved, while an unsuccessful startup destroys the value of those involved. Since the outcome is almost binary, the only way to spread the risk of the outcome is through mass. VCs can invest in many start-ups in order to reach the moving expected value in the binary expected value, while founders can only manage one start-up, which is why the binary expected value approaches zero. This leads to the fundamental conflict of interest in startup financing, which will be discussed in another paper.
Conclusions for practice
To believe that Lean Startup can be expected to build a successful business is a fallacy. A founder cannot assume that he or she will get a good financial return from it. Founders need to pay themselves a decent salary and take care to compensate financially for the risk. Since investors clearly do not have this interest, we have to appeal to them to support the founders in the long run.
Furthermore, the mathematical model shows that, in addition to the Lean Startup for finding local optima, we need a methodology to be able to minimise risk globally. This is the task of research. I would be happy to give interested researchers a deeper insight into the topic.
Where does the data come from?
We conducted this study at HHL as part of a statistical master's thesis. The data comes from publicly available time series from US government agencies. Accordingly, our data base is limited to US companies. Our time series runs from 1990 to 2020, and is corrected for many influencing factors, such as economic cycles.
Comentarios