Detecting and Reducing Interruptions at Work
Supervisors: Denis Lalanne, Remy Zimmermann
Student: Thibaut Mauron
Project status: Finished
It has been shown that human interruptions are one of the many factors affecting workflow and productivity. Our research question is: can an interactive device with ambient visualization lower the number of human interruptions when workers do not want to be interrupted? To answer this question, we started with a study involving sixty workers in order to define their needs. Based on that study we built a first prototype which was tested in-vivo for one week with five users. The result of that first iteration led us to the second, more advanced, version of the prototype. This final prototype is capable of sensing interruptions, informing co-workers that the user does not want to be interrupted, and it gives an estimation of the remaining busy-duration. We manufactured five instances of our second prototype which were tested in-vivo over two weeks.
To conduct proper testing, we added sensing capabilities to our devices in order to collect quantitative data. In parallel, we interviewed the users daily to gather qualitative data.
This study is about social relations and behaviour at work, but there is also an important technical part including inter-device communication protocols, design of 3D-printed mechanical hardware parts, different programming languages, sensor data analysis, multi-threading, and the use of different web services.
In total, we conducted a study involving multiple surveys with up to ninety participants, two design-implementation-test iterations of a prototype, and a two-week long experiment in-vivo with five users in a 300-employee tech company. This article discusses the design of the devices based on survey results, the hardware and software parts required to make it work and communicate with the cloud, the experiment, and the evaluation procedure (both qualitative and quantitative), as well as the limitations of the study.
The analysis of the results shows that it is possible to automatically sense interruptions, and that the device has a positive impact on the number of interruptions when people do not want to be interrupted. In addition, we have shown, that the device is more efficient if it is used for several short amounts of time (e.g., 30 minutes), rather than for a long period (e.g., full day).
Keywords: Interruption, Flow, Productivity, Time Management, Sensing Proximity, Analyse Employment, Software, Hardware, Cloud, AWS