CLIENTUnited Arab Emirates
DATEFeb 8, 2019
Technology UsedQuBole, Java, Python, Tableau,
A DTH-OTT provider from the Middle East wanted to power up its customer support function by streamlining credits distribution and disbursement as per the consumer’s media consumption trends. The prevalent trends entailed a human customer support executive disbursing credits in order to enhance customer loyalty and give a thrust to the overall customer experience. However, the foundational flaw with the system was the arbitrary manner in which credits were being distributed by the agents to the customer. This was one of the reasons that saw the DTH operator facing substantial customer churn. The operator needed a suitable intervention that can prevent the unwanted distribution of credit without compromising the level of customer satisfaction that the media consumer was availing. They were looking to streamline the distribution of credits and prevent customer churn from happening due to misaligned credits assigned to a given consumer. Along with this, they also needed assistance in getting better value from the payments collection efforts put in by the agents. A bigger issue was with maintaining customer relationships that could avoid consumer moving to competition. This was driven by the need for an optimization of the credit distribution process with a data driven system.
At AdFolks, the mandate was clear from the DTH operator – to enhance the overall customer experience by deploying a seamless and streamlined credit distribution system. This meant that a system needed to be in place that automatically factored in the consumer state of mind and allocated credits accordingly in order to bring down customer churn. The solution also helped the DTH operator to gain insights about which segments witnessed a greater churn so that customer relationships can be strengthened and customer experiences can be improved.
Our solution tackles three key areas in addressing the client issues – 1 - Consistency in credits disbursement We first analyzed subscription, call center, payment and other customer interaction data to carry out customer and agent segmentation. It was then correlated with the credit offered by agents to measure the impact on customer behavior. We came up with a recommendation engine that gives out suggestions on the apt credit value to be allocated to a particular customer. We used regression analysis for this purpose. We also highlighted anomalies in credits versus customer profile to improve customer support team’s performance. 2 – Churn Score We came up with a churn score based on vital lagging indicators. The score (between 0 and 1) could detect likely churn and customer loyalty at an individual consumer level with 80% precision. 3 – Payment reminders personalization A key finding of our initial research was the different times at which an individual consumer would make payments for the services. Since the automated payment reminder calls and communications were pre-programmed, there was no provision to personalize the payment reminders based on an individual’s historical data on past payments made. We employed machine learning to learn the correlation between collection attempts data and subscriber historical details. As the model keeps ingesting bigger volumes of data, the precision continues improving significantly.
The solution has driven advantages at three key levels to boost overall customer experience. 1 – Better and automated credits disbursement Instead of random credit distribution, the solution provides better avenues to correlate customer frame of mind and tie it up with the appropriate credits to be allocated. As an outcome, the right volume of credits can be provided based on appropriate customer behavior in real-time. This is evolving soon into a system that can automatically award credits with no human involvement. 2 – Better churn management With the supervised classification model, the client is able to better forecast churn before it actually happens. This gives the operator the time and opportunity to improve customer relations before consumer attrition to a competitor brand. 3 – Bespoke payment reminder system We have deployed personalized recommendation of payment method and payment date. This not only improves the CX but also the efficiency of the collections team with better suggestions on when to remind an individual consumer.