Across industries, customers increasingly expect interactions with companies to be simple, predictive, and seamless across channels. Three examples demonstrate this kind of enhanced customer experience:
- A North American utility proactively notifies customers about higher bills during the winter and provides customers with tips on how to reduce their bills.
- A bank alerts its customers about potentially fraudulent transactions and can replace debit cards within 24 hours.
- An airline provides a seamless customer experience when customers move from its app to interactive voice response (IVR) to an agent.
The common thread in these examples is the role that advanced analytics plays in enabling exceptional customer service across channels. These elements can improve customer satisfaction (an increase of 10 to 20 percent) while unlocking lower cost to serve (cost savings of 20 to 30 percent in some cases). For these reasons, companies are seeking to step up their customer service offerings to meet the heightened expectations of customers.
Capturing cost savings and improving satisfaction don’t necessarily require complex use cases. Targeted use cases, such as speech analytics, can boost performance by 10 to 20 percent by enabling more effective agent coaching. And IVR analytics can improve containment rates by 5 to 10 percent.
Because many utilities have yet to implement advanced analytics in customer care at an enterprise level, we typically recommend proceeding with a two-phase approach. Utilities should first ensure they have a robust foundation of organizational capabilities and acumen. Then they can explore advanced techniques that enable next-generation use cases to unlock greater levels of cost savings and improved customer satisfaction.
Phase one: Laying the foundation
To begin integrating advanced analytics, utilities first need to identify and prioritize use cases, aggregate the necessary data, establish cross-functional teams, and adapt operations.
Identify and prioritize use cases
Multiple existing advanced-analytics applications have the potential to improve customer service and productivity. But different use cases do not generate equal value, and without guidance, analytics teams typically will not know which use cases to prioritize. To avoid leaving significant value on the table, utilities should ensure their analytics teams work closely with the business to assess a list of use cases based on the following criteria:
Potential value. Estimating the value of a use case should include considerations such as a clear view of the baseline, the value at stake, implementation costs, and the time to impact.
Technical readiness. The organization’s technical maturity—encompassing elements such as the availability of data and resources, deployment considerations, the likelihood of adoption by end users, and scalability—can dictate how quickly a use case can generate value.
Strategic fit. A use case should be aligned with the organization’s overall goals, particularly regarding its potential to improve customer satisfaction.
Aggregate data for prioritized use cases
Disconnected data sources often lead to inaccurate or incomplete insights. For example, the combination of data from the customer relationship management (CRM) and automated call distribution systems can unlock a host of use cases, including “call propensity by customer segmentation.” Utilities can gain a tremendous advantage by harnessing the power of all available data sources for prioritized use cases, thus serving customers more effectively. They must also develop the ability to connect data from different sources using unique customer identifiers.
Utilities should start with readily available data sources to prove the value before adding other data sources to strengthen the use cases. For example, a customer journey use case would require aggregating customer data from multiple channels, identifying the break points, and using these insights to design future interventions. One utility that embarked on this use case augmented contact-center data with data from interactions across the web, mobile, and IVR (Exhibit 1). Analysis revealed more than 50 percent of the customers were not using digital channels prior to reaching out to the contact center, with app usage particularly low. These insights led the utility to focus on increasing customer awareness of its digital channels and revamping some existing digital functionality, such as simplifying bill formats, using visuals to show bill trends, and reducing the number of steps needed for a customer to start a service.
Build a cross-functional team
Many analytics teams work in silos (such as an analytics center of excellence), which means they lack an understanding of the organization’s greatest pain points and miss out on receiving regular feedback from the business on the applicability of the insights they generate. For example, to get the most from analytics-driven models that draw on employee data, the analytics team would need feedback from the managers in charge of hiring, recruitment, and operations. Close collaboration among multiple teams is critical.
One US utility was able to identify and fix common customer break points in its IVR by promoting a collaborative approach. The analytics team worked with the business to identify customer drop points and closely coordinated with the IT team to make quick adjustments to the IVR. The team then tracked the impact of these changes on customer behavior. The analytics team was able to gauge how frequently customers used self-service features to identify the most- and least-popular nodes and reorder the options accordingly (Exhibit 2). Most were related to problems with technology workflows on the back end or customers being confused by the verbal prompts in certain nodes. By itself, the analytics team could not have identified such adjustments.
Capturing value from customer-care use cases requires businesses to make appropriate operational changes. For example, increased digital adoption through targeted customer education would decrease the volume of live calls to the contact center, reducing the number of frontline staff required. Operations should coordinate with the workforce management team to ensure the team revises staffing projections to capture the total value of the use case. Analytics can also play an essential role in tracking KPIs to ensure the use case is generating the anticipated value from initial estimates.
Phase two: Next-generation use cases
Once utilities have mastered the foundational elements of analytics, they should prioritize the next generation of use cases. This process typically requires brainstorming with the business to come up with a list of new use cases, capturing incremental data such as text transcription data from voice recordings, and enhancing their understanding of advanced-analytics techniques. We discuss below two examples of these techniques: machine learning (specifically to build predictive-intent models) and natural-language processing (specifically for speech analytics).
Once utilities have mastered the foundational elements of analytics, they should prioritize the next generation of use cases.
Anticipating why a customer might call is not only a key to excellent service but also the future of customer care. To gain this capability, utilities will need to build algorithms that use advanced machine-learning techniques and data from previous interactions across channels (such as email, chat, apps, and IVR) to forecast call intent.
For example, one utility deployed a predictive-intent model to reduce calls related to unexpectedly high bills. The algorithm estimates the individual customer’s upcoming bill and calculates the deviation of the bill from the amount billed the previous month and year. If the deviation is significant, the utility proactively sends a notification to the customer and explains why the upcoming bill could be higher. Similarly, we have also seen predictive-intent models used to adapt IVR to proactively resolve customer concerns.
Today, these models commonly deliver meaningful value across industries outside the utility space. For example, airlines predict call types based on the customer’s most recent communication and queries from previous calls; if the purpose of a customer’s past few calls involved time of departure, that option is listed first in the IVR. Utilities can learn from successes in other industries.
The combination of good call recording, advanced speech-to-text conversion, and text analytics (another use of machine-learning algorithms) can generate extensive call-based insights. Specifically, algorithms can pinpoint additional opportunities for call resolution, and speech analytics can support use cases such as agent coaching, call drivers, customer satisfaction drivers, and the automation of quality assurance.
One utility built a speech analytics use case to identify agent-coaching needs by call types. The traditional approach was much more manual, with a handful of supervisors listening in on calls. By harnessing analytics-driven insights, supervisors could understand agent needs at a much more granular level and engage with agents more meaningfully in coaching sessions (Exhibit 3).
Many utilities aren’t capturing the full value of advanced analytics today due to a lack of prioritization, limited ability to harness multiple data sources, and limited collaboration across multiple departments. Developing these capabilities will be critical for any utility that expects to be a customer experience leader and provide better service at a lower cost.