Let’s start with the definition of a startup:
Startup: novelty technological company with limited resources that operates in a high uncertainty environment.
The problem with the uncertainty is that it is worthless to have a business plan for 5, 3 or even 1 year. If we add the fact that the startups don’t have lot of resources, it is very probably that most of the startups will fail. But not all; there were several startups that have created great products that nobody expected to be a great success. Some of them are now one of the biggest companies in the world, just because they risked it with a new technology or product that nobody knew how the people could use it.
What is the secret for a startup to succeed? The only way to deal with the uncertainty and create a successful company is to learn quickly. Without more information we can think that the goal is to launch a minimum viable product and see how the market behaves. Then we can continue launching new versions of the product and keep watching what happens. We could even have growth as this graph shows:
At first sight it could seem that the company is doing well because it is getting more and more paid users. Nevertheless, in this case we are wrong. If we see the percentage of paid users related with the visits, (visits are the 100%) the graph is the next one:
As we see the percentage of paid users is decreasing. It is possible that the growth in the number of paid users is a consequence of a big investment in marketing that is generating more traffic to the website, but sooner or later the sales will collapse and nobody will know why. The company will be stuck in the land of the living dead companies; it would get a moderate success but won´t be able to grow anymore. Here is when the scientific method is important:
Scientific method: Set of steps set in advance in order to achieve valid knowledge through reliable instruments, standard sequence to ask and answer a question.
When the minimum viable product is launched to the market it has to be to answer an initial hypothesis. Without wondering the proper questions we are going to learn nothing.
When you test to see what will happen you will always succeed, because you always see that something happens, although nothing is happening.
It is very important to choose the right drivers that lead us to a good understanding of what is happening in order to answer our initial question. With drivers like the ones of the first graph, we would trick ourselves and feel that the company is growing well, when this is false.
The process should be the following:
- Ask a hypothesis
- Choose the correct drivers to answer the hypothesis
- Launch a minimum viable product in order to check the hypothesis
- Analyze the results with the user feedback and the drivers
- Create a new hypothesis and star over again
This cycle allows us to learn, but it also should be done quickly. The faster we are learning and validating the hypothesis the faster we respond to the needs of the users and the less resources we spent. Furthermore, we won´t spend time developing features that are not going to be valued by the customers.
This method is valid when the company is starting and growing because the impact over the number of users is minimum. But when the company has grown and has a known brand it is not recommendable to experiment with all the users. It could lead to a great insatisfaction and decrease the sells. The impact of the experiments has to be controlled and it has to be launched only in a few customers, just to create the minimum negative impact.