Delivering what was promised: An action research in a journey of a software startup Akkadu in making realistic commitments based on reliable and metrics-driven estimations

Reliable software delivery is a pervasive problem in the software industry. However, there is a lack of research on the topic in the startup sphere. This is the problem a Beijing based startup Akkadu faced in early 2018. This study seeks to answer the research question of "What kind of software development process will serve the team in tracking development progress and forecasting completion dates?". As a result, this research also seeks to help fill in the lack of startup-focused software process research. The methodology used is action research which includes iterative planning, acting, and fact-finding steps. These are supported by situation analysis and a literary research on Scrum, Kanban, and Scrumban. The startup environment seems resistant to software process. Selling software process improvement as a trial seems like an effective strategy to combat this. Relevant metrics to visualize progress seemed to keep the team engaged untill process was internalized. At Akkadu this took eight Sprints. The results of this research indicate that a story point-based estimation can provide accurate estimations for a Sprint scope. It seems that higher average, and individual, story point counts increase the likelihood of failed delivery. Refining items and standardizing the backlog can provide improvements, as long as it produces smaller backlog items. Over time, story point sizes and the number of items completed in a Sprint seem to standardize. This could indicate that story point estimates could eventially be replaced with item counts. It seems that a startup environment can benefit from software process improvements without sacrificing agility. The total work output of the team declined during the research, but this did not reduce feature development speed. Systematic refinement and prioritization lead to better work quality. Consistent improvements started at Sprint eight, which indicates that team buy-in could be key in succesfull software process improvement