Companies that use robotic process automation very often deal with problems maintaining the bots in their RPA portfolio. Current key performance indicators do not track the behavior of RPA bots or processes. For better maintainability of RPA bots, it is crucial to easily identify problematic behavior in RPA bots. Therefore, we propose a strategy that tracks and measures the behavior of processes to increase the maintainability of RPA bots. We selected indicators of statistical dispersion for measuring variability to analyze the behavior of RPA bots. We analyzed how well statistical dispersion can describe the behavior of RPA bots on 12 processes. The results provide evidence that, by using statistical dispersion for behavioral analysis, the unwanted behavior of RPA bots can be described. Our results showed that statistical dispersion can describe the success rate with a correlation of -0.91 and outliers in the data with a correlation of 0.42. Also, the results demonstrate that the outliers do not influence the success rate of RPA bots. This research implies that we can describe the behavior of RPA bots with variable analysis. Furthermore, with high probability, it can also be used for analyzing other processes, as a tool for gaining insights into performance and as a benchmark tool for comparing or selecting a process to rework.

翻译：采用机器人流程自动化的公司常常面临如何在机器人流程自动化（RPA）组合中维护机器人的问题。当前的关键绩效指标无法追踪RPA机器人或流程的行为。为了提高RPA机器人的可维护性，轻松识别其问题行为至关重要。因此，我们提出了一种策略，通过追踪和度量流程行为来增强RPA机器人的可维护性。我们选取统计离散指标来衡量变异性，以分析RPA机器人的行为。我们分析了统计离散性如何描述12个流程中RPA机器人的行为。结果表明，通过使用统计离散性进行行为分析，可以描述RPA机器人的非期望行为。我们的结果显示，统计离散性可以描述成功率，相关系数为-0.91，描述数据中的异常值，相关系数为0.42。此外，结果还表明异常值并不影响RPA机器人的成功率。这项研究意味着我们可以通过变量分析描述RPA机器人的行为。而且，该方法很可能还能用于分析其他流程，作为洞察性能的工具，以及作为比较或选择需重新处理流程的基准工具。