How Digital Service Providers (DSPs) can Build AI-ML Capabilities to Enhance Efficiency of Technicians and Reduce MTTR by 30%

This Industry Viewpoint was authored by Boominathan S and Abhay Goyal of Prodapt

Inefficiencies in the field service are a major challenge for most Digital Service Providers (DSPs). Delay in providing fixes impacts customer experience leading to significant customer churn and revenue loss affecting profit margin for DSPs.

Traditionally, 70% of the field technicians are dispatched to the field without any insights about the right issue, fault location, nature of the problem & solution recommendations. This results in an extended time to resolve the issue, partial fixes, and repeat dispatches. Over 30% of repeat dispatches occur due to improper fault isolation and 47% of dispatches require rework/repeat dispatch due to inappropriate actions taken by field technicians.

Field technician efficiency, if not improved has a cascading impact such as:

• Increase in the number of repeat dispatches
• More time to resolve customer issues
• Higher cost to the DSPs
• Higher customer churn rate due to low FTFR (First Time Fix Rate)

According to Gartner’s report on “Critical Capabilities for Field Service Management,”  70% of customer interactions will involve an emerging technology such as machine learning applications, chatbots, or mobile messaging, by 2022.

To create a great customer experience, DSPs need to build and implement an AI-powered field service framework that helps technicians with the right fault location and guided actions as stated below:

 Fig1. AI-powered field service framework to improve First Time Fix Rate (FTFR) and field tech’s efficiency

Figure-1 illustrates 3 key components of the AI-powered field service framework. 1. Fault location classifier – ML model pulls out data from devices/systems to perform multivariate analysis and identifies the right fault location. 2. Recommendation engine – ML model helps with guided actions to resolve the issues quickly. 3. Further, the technician dashboard helps the technician in better planning by providing a one-stop view for all the productive dispatches assigned to him/her.

How to build key components of the AI-powered field service framework?

In this section, we will show you how to implement the AI-powered field service framework and the key components that need to be built for enhancing the field technicians’ efficiency.

1. Fault Location Classifier

One of the major challenges in the traditional approach is that the field technicians spend 60%-70% of the time isolating the issue. Most of the time, the technician starts the troubleshooting at the wrong location. The fault location classifier – ML model can predict the right fault location.

Why ML model? – The complexity involved is high. We need to pull out data from 40+ devices/systems, each having 100s of distinct parameters. Hence, the need for an ML model is imperative as shown below.

ML model enables the technician to visit the right fault location at the first time itself. Building fault location classifier – ML model helps DSPs in reducing 46% of the time taken to find the fault location & thereby improves the NPS by 20%-25%.

2. Recommendation Engine

In the traditional approach, the technicians spend 66% of the time identifying the right fix for the issue.

Also, 20% of the time is spent analyzing if the fix is correct, post fixing the issue. Recommendation engine – ML model can recommend the guided actions to the technician to resolve the issue quickly.

Need for ML Model – Along with pulling out data from devices/systems, the model also considers the below parameters to predict the next best action:

• n number of problem types
• n number of ways to resolve the issues
• fault at n number of locations

The model collects data for all the parameters from all possible sources and the right fault location to:

• Perform text analytics on historical chat data to identify a correlation between issues-fixes and then train the model
• Classify if the issue is due to weather extremities using weather data like wind speed, precipitations and recommends postponing the dispatch
• Recommend on the dispatch tickets for mass outages based on ZIP code and DSLAM level aggregation
• Retrain the model by capturing all the feedback/remarks from technicians and improve the accuracy of the model

Implementing a recommendation engine ML model can save 50% of time spent in resolving issues, thereby improving field technicians’ efficiency by 30-40% and saving OpEx for DSPs.

3. Technician Dashboard

Technician dashboard – Provides a one-stop view of all the productive dispatches to a technician in real-time with fault location and guided recommendations.

Fig. Sample technician dashboard for a specific technician

Technician dashboard helps in showcasing AI/ML model outputs such as right fault location & guided actions to resolve the issue. It provides a one-stop view of all the productive dispatches assigned to the technician for better planning and capturing remarks for an improved feedback mechanism.

In conclusion…

With the AI-powered field service framework presented in this article, DSPs can improve field tech’s efficiency and customer delight. Implementing the above 3 components can help DSPs to reduce the MTTR (Mean Time to Resolve) by 30%, improve FTFR (First Time Fix Rate) by 25-35%, and save OpEx by 30%.

Authors

• Boominathan S – Director, Delivery, Prodapt
• Abhay Goyal – Senior Analyst, Strategic Insights, Prodapt

If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!

Categories: Artificial Intelligence · Industry Viewpoint · SDN