Planned maintenance is a daily activity for any number of complex systems, including cable plants. It is important to think of a cable plant as a living, breathing organism that requires care and feeding, involving replacing parts that are continuously exposed to the elements. Repairs often include identifying problems, making repairs, and replacing parts while temporarily interrupting a customer’s service. A service interruption event (SIE) averages between five to ten minutes. In an ideal world, all SIEs would be performed during the evening maintenance window, but in practice, most short-duration SIEs must be performed outside of this maintenance window. Currently, SIEs are scheduled without the benefit of knowing which hours would have the least or the highest amount of subscriber impact. This information would be invaluable in optimizing the ideal time to perform an SIE.
We set out to find if a data-driven system could be developed to determine the best time to conduct SIEs. Performing SIEs during times when they will have the least impact on subscribers would not only provide a better subscriber experience but also could potentially cut the expenses incurred by responding to customer interactions (CI), such as calls to our care agents, unnecessary truck rolls, chat sessions, and other triaging events.