Throughout the project period of six months, we were able to verify the technical feasibility of these features by collecting the required data, cleaning and enriching the data, and finally building the predictive models with an acceptable model fit. Simultaneously, we validated the market demand further and deployed pilot studies to ensure customer-centric product development. The key learnings comprised: 1) The relationship between startups and incubators is burdened in many cases. Therefore, the startups need to be in the driving seat when using our software. Data should be collected from their progress and daily activities to make the process as least intrusive as possible. 2) Our initial startup interface did not suffice the requirements of the startups, which prohibited implementation. Consequently, we developed an interface based on the startups’ needs. 3) Given their inexperience and lack of founding knowledge, startups have difficulties in setting goals and defining concrete plans. They need substantially more guidance and training on the fly as it provided by incubators. 4) We learned to separate incubators between professionally managed high-touch incubators and poorly managed low-touch incubators. This helped us to narrow down the target group.