Dell Technologies Ltd - Innovation in Customer Service Management, Planning & Practice

Company: Dell Technologies Ltd., Bangalore, Karnataka, India
Category: Award for Innovation in Customer Service Management, Planning & Practice > Computer Industries

Nomination Title: Customer Experience Prediction Index

Essay:

As a customer centric organization Dell has been striving hard to accomplish our mission of ZERO Customer Dissatisfaction. The diversified usage of computer software roll into 500+ major issue types which multiplied with hundreds of hardware customizations make landscape of technical support very complex and to ensure that defects are minimum on same, at dell we have created a robust and comprehensive enablement system for our front-line agents.

Dell’s tools and knowledge management ensure support system is very efficient however owing to numerous resolution path possibilities and human bias we still observe a few micro defects and hence it is very important that each customer journey is watched for any aberration which in human capacity is not possible at all. Hence using technology as enabler, Dell has been working on a prediction model which can precisely identify critical customer journeys so that an in-time service recovery can be attempted on those customer journeys directed as prescription by triggers classified in concerned customer journey.

Dell studied defect reviews as well as proactive reviews and mapped same into a structured process to identify all customer journey features that compose these reviews and created an integrated engine which can holistically map all aspects of customer journey. This Engine was shaped into an Automated Intelligence engine enabled by Machine Learning particularly Classification Algorithms, Text Mining & CRM Analytics based on business rules. Using this engine, a pool of 100k customer journeys was created which had customer output available and hence have been able to create a correlation model between customer journey features and customer satisfaction levels. This has enabled us to assign a discrete value system to these customer journey features and engineer a quantification system which can assign weightage and impact to all these customer journey features as seen in correlation of these to customer satisfaction levels.

In a nut shell, Dell has been able to create an automated mapping system to create customer journey features, demystify how different customer journey attributes impact customer satisfaction level and put numbers to it so as to create an Index which is directly a probability score of satisfaction of a customer. This has been christened as Closure Recommendation Index (CRI) and was embraced as first phase of DSAT Predictor. This prediction intelligence in the form of index is available to business so that users can attempt service recovery on low index customer journeys and ensure Time Taken to Closure (TTC) is minimal on all high index customer journeys. This intervention mechanism targets all the aspects of contact center impact on customer journeys, thus has a potential to reduce contact center defects in customer dis-satisfaction by 40% as estimated through HBR Analytics Research*

While prediction accuracy of more than 95% through CRI on Satisfaction prediction has been attained, when prediction modelling on Dissatisfaction was studied – it was found that pattern detection on dissatisfied customer journeys will need a structural change in approach so NLP in addition to conventional Machine Learning Classification model was introduced for more preciseness of customer journey features and for probability indexing, decision tree was utilized to boost earlier used SVM (support vector system) along with hyper-tuning. Using similar concept as used in Closure Recommendation Index (CRI) boosted with evolved analytics techniques, a probability index for predicting Dis-satisfaction was created, namely Distress Intervention Index (DII).

For effective utilization and single source user execution we have merged CRI and DII to create Customer Experience Prediction Index Users access a portal which is updated every 6 hours with all the customer journeys impacted in last 6-hour span and index segregation is available on all cases along with journey feature value (which act as directional prescription)