Improving Feature Lead Time in Digital Service Development
Saastamoinen, Toni (2023)
Saastamoinen, Toni
2023
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2023052614868
https://urn.fi/URN:NBN:fi:amk-2023052614868
Tiivistelmä
The case company has a decades-long history as a digital frontrunner in its industry. To remain competitive in the future as well, it has set ambitious targets for digitalization. While investing in digital leadership, income growth and operational efficiency are also strong focus areas. The case organization plays a significant role in realizing these aspirations while maintaining strict cost control. In an ever-tightening competitive environment, it is crucial to speed up the time to market, i.e. feature lead time when developing new digital services.
Research strategy for this project was case study. The case organization is a SAFe based large Agile Release Train (ART) responsible for number of value streams, having a need to shorten the feature lead times of its digital deliveries. Current baselines for feature lead times were established at first on ART, development team and selected value stream levels, in order to be able evaluate the effectiveness of the improvement measures in the future. This was done by using statistical analysis on historical feature lead time data. After establishing the baselines, the next step was to investigate the reasons for long feature lead times in selected value streams. That turned out to be impossible to perform just based on the lead time data. Quality of the data was not reliable and comprehensive enough for conclusive analysis. Therefore, manual in-depth analysis was performed on features with a very long lead time, in this case more than one year.
Analysis revealed two main reasons and few smaller factors for long feature lead times. In many cases the work items were too large to begin with, which has led to slower implementation, longer lead times and thus delayed value realization. Teams have often failed to address the delays actively and swiftly when they were noticed. Introduction of new technology tends to lead into increased lead times. The same happens when teams fail to identify and mitigate risks in advance related to complex features. Analysis exposed problems with the lead time data quality. It is sufficient for measuring overall feature lead times, but not reliable enough for more detailed analysis of the feature development and delivery process.
This study was able to identify number of root causes for long feature lead times. The study also produced a list of suggested measures for the case organization, not only to improve feature lead times in the future, but also the quality of the lead time data.
Research strategy for this project was case study. The case organization is a SAFe based large Agile Release Train (ART) responsible for number of value streams, having a need to shorten the feature lead times of its digital deliveries. Current baselines for feature lead times were established at first on ART, development team and selected value stream levels, in order to be able evaluate the effectiveness of the improvement measures in the future. This was done by using statistical analysis on historical feature lead time data. After establishing the baselines, the next step was to investigate the reasons for long feature lead times in selected value streams. That turned out to be impossible to perform just based on the lead time data. Quality of the data was not reliable and comprehensive enough for conclusive analysis. Therefore, manual in-depth analysis was performed on features with a very long lead time, in this case more than one year.
Analysis revealed two main reasons and few smaller factors for long feature lead times. In many cases the work items were too large to begin with, which has led to slower implementation, longer lead times and thus delayed value realization. Teams have often failed to address the delays actively and swiftly when they were noticed. Introduction of new technology tends to lead into increased lead times. The same happens when teams fail to identify and mitigate risks in advance related to complex features. Analysis exposed problems with the lead time data quality. It is sufficient for measuring overall feature lead times, but not reliable enough for more detailed analysis of the feature development and delivery process.
This study was able to identify number of root causes for long feature lead times. The study also produced a list of suggested measures for the case organization, not only to improve feature lead times in the future, but also the quality of the lead time data.