Prediction 2.jpg

The popularity of Customer Experience (CX) as a new mantra of business has resulted in a tidal wave of feedback programs. From a consumer standpoint nearly every transaction in our lives is followed by a survey. This preponderance of surveys has the potential to negatively affect customer experience.


To counter this, Customer Experience programs need to move beyond surveys and innovate intelligent ways to evaluate customer satisfaction and loyalty. At EMC, we have begun to use our collected CX feedback to predict customer satisfaction without the need for a survey. Advancements in big data and statistical analysis tools make it possible to identify impact drivers and trigger points of satisfaction at a fine level of granularity from collected survey data. By applying this analysis against UN-surveyed events, EMC can reasonably predict customer satisfaction, or dissatisfaction, without the need for direct customer feedback.


At this time, EMC receives survey responses from only 1.2% of all events handled by our Customer Services organization. From the analysis of these responses, we are able to predict when dissatisfaction may occur in the other 98.8% of activities. The Predictive Insights Program is designed to address this broad potential for dissatisfaction.


The long-term goals of the Predictive Insights Program are to:


  • Reduce the frequency and volume of EMC’s survey programs;
  • Extend the reach of captured survey data by projecting results onto events that are not surveyed; and
  • Take action on predicted dissatisfaction.


The ultimate objective of the Predictive Insights Program is to embed the statistical prediction model directly into EMC’s call/ticket
management system to identify and address issues in service delivery in real-time and proactively before the issues generate customer dissatisfaction.


Since deployment, we’ve measured the accuracy of the prediction model to be approximately 65%. In other words, 65% of the time that
the model predicts that a customer may be dissatisfied, EMC Customer Services did, in fact, close the event with the customer feeling a level of dissatisfaction with the services received. It is the goal of EMC’s CX team to reach a 75% accuracy level before the model is embedded into our call system.


Once dissatisfaction is predicted, the responsible Service Manager for that event is systematically notified and is required to follow-up
with the customer to identify and address any dissatisfaction.


Currently, the Predictive Insights Program is deployed within a subset of service groups that have historically underperformed in
satisfaction metrics. The results are showing great promise. For example:


  • Up to 79% of participating Service Managers indicate that the predictive follow-up improved the experience of customers;
  • The improvement in the customer satisfaction levels within participating organizations is growing 50% faster than non-participating organizations;
  • Up to 14% of predictive follow-up calls result in re-opening of the original ticket because the customer was not satisfied with the resolution;
  • Up to 4% of follow-up calls result in new business opportunities (e.g., new or renewed maintenance contracts, new equipment sales);
  • The number of responses to the Customer Services transaction survey for participating service groups has increased 180% more than non-participating groups as customers learn we take action on feedback; and
  • Satisfaction with Customer Services measured in EMC’s relationship survey has increased 2% overall since the deployment of the program.


Overall, the Predictive Insights Program has resulted in a significant increase in ROI, or Return on Information, of collected CX data. Through this action-based predictive program EMC can extend its reach to more customers and, at the same time, reduce the current trend towards survey proliferation. Our “prediction” for future benefits looks strong.


Take care,

Brad Barker

Consultant Customer Advocate
Total Customer Experience

EMC Global Services

Twitter: @BradB555