Contributed by: Pushpak

I am a certified PMI-ACP professional based out of Kolkata & currently working as a Global Manager in Telecom Industry, where I am responsible for ensuring customer satisfaction for some of the largest Tier I and Tier II Communication Service Providers around the world.

My role at present is to manage the Customer Support team for the Optimization LOB. I was tasked to setup the Support process and ensure that all tickets raised by the different tier-I and tier-II operators were resolved within the SLA. 

For a Customer Support team, different customers have different SLAs, based on their contractual agreements and other business parameters. Hence the target is to never miss an SLA. Likewise, the technical SMEs working on the tickets also have different expertise level, different communication skills etc.  

Hence, it always makes sense to map the most critical issues, or the most premier customers to the appropriate SMEs. All this while, it was done based on intuition of the team, was person dependent & sometimes error prone.

After analysing the process, I found that there are certain types of tickets which are difficult to understand, hence complex to resolve, thus taking more time & leading to breach of SLAs. Missing SLA had severe business implications like customer escalations, loss of goodwill, and in worst case, penalty. Hence, we always wanted to avoid it at all costs. 

Similarly, providing visibility to the customer regarding issue resolution timelines helps increase customer confidence. 

To come up to a solution to this problem, I did Exploratory Data Analysis to understand the patterns in ticket resolution. Also, Hierarchical Clustering was used to group the different category of tickets. And finally, using Linear Regression, I created a model which would help predict by when the open issues could be completed.  

These Analysis generated very useful insights where we were able to capture :

  1. Top 10 customers by number of complaints – these were our focus customers as they faced multiple problems
  2. Mapping of Issue category to customers further helped us in understanding the type of issues faced by each customer, as well as the opportunities and threats 
  3. Customer wise ticket ageing with Product category as hue showed some issues taking longer time to be solved in comparison to others.
  4. SME vs Product wise ageing to understand the strength of each person
  5. Using Hierarchical Clustering, we could see that 3 distinct category of issues:
    1. 1570 were easy and got resolved fast
    2. 150 with medium complexity
    3. 5 which missed SLA             
  1. Linear Regression was finally used to come up to a model for prediction of issue resolution timelines. The equation which was arrived at is:

The solution/recommendation proposed to solve the business problem:

From both the clustering methodologies, it becomes clear that some tickets (around 1%) missed their SLA (DaysToExpire is negative). This falls within the acceptable tolerance of the company. However, most of the tickets get resolved in about 2.72 days time with about 13 days SLA still left. This is good performance.

However, we find in about 150 tickets that the Ageing is high. This means that the solution had been provided to the customer, but the ticket could not be closed as the solution was not implemented by the customer.

This can reflect a lack of confidence of the customer in implementing the changes. This means either the SME needs to better explain the changes to the customer to increase his confidence, or implement the changes in a demo environment and show the results to the customer before implementing in production.

The Linear model was helpful in prediction of timelines. This helps in increasing the customer confidence. 

Using statistical analysis for Support teams is quite new to the organization. The EDA charts have given us very nice visibility of different aspects of the team performance. The management is quite impressed with such analytics as this will help in overall customer satisfaction improvement. 

In parallel, for every ticket now being raised by the customer, we have made it a process to update the customer about the ETA for the solution. 

Lastly, this activity has opened up new avenues for me. Apart from several appreciations from management, I have also been tasked with implementing such analytics in the other LOB Support Teams. The Global Management is even thinking of creating a new department eventually for Business Analytics – when that happens, I will be one of the founding members of the team. 

Inspired by this story and wish to upskill? Join Great Learning’s PGP Data Science and Business Analytics Course. For more such success stories, watch this space.



Please enter your comment!
Please enter your name here

4 × 2 =