Experts from seven companies share their thoughts on whether real-time and predictive analytics are a valuable addition to contact centres and what human agents can learn from these analytics
Real-time and predictive analytics can be a powerful tool in delivering exceptional customer experience. By making use of these types of analytics, contact centre agents are able to smooth out any pain points of the customer journey while simultaneously providing an empathetic experience and reducing wait time.
In this month’s edition of the CX Today round table, we welcome:
Experts from seven companies share their thoughts on whether real-time and predictive analytics are a valuable addition to contact centres and what human agents can learn from these analytics. Here’s what they had to say.
How can real-time and predictive analytics impact the customer journey?
Ziv: “The role of real-time and predictive analytics is to anticipate and provide the right answers to customers. This can be done via self-service journeys and in the agent-assisted journey, but both need to be informed by data via recent customer interactions.
“For many years, self-service had a bad reputation; it was a gatekeeper that the customer had to get past. Today, real-time and predictive analytics infused with IVRs and IVAs are reinventing self-serve to be more relevant, intuitive and effective.
“With agent-assisted journeys, the goal is to leverage real-time analytics to anticipate issues and concerns and to help the agent better serve the customer. Verint’s approach leverages and infuses triggers from three unique sources to guide employees in real time: Linguistic — recognizing specific words, or a positive or negative sentiment, Acoustic — recognizing sound triggers, such as long silences and interruptions, and Application — adding critical context and recognizing actions from the employee’s desktop, such as adherence to processes.
“Employees are notified of recommended actions to take in a unified Work Assist application, so they can immediately guide the call to a better outcome. Predictive analytics helps build these models to be more effective over time.”
Einstein: “AI predictive analytics is especially useful when mapping customer journeys because the AI finds patterns and begins to predict not only where a specific journey can be improved, but also offer prescriptive next-best-action for mitigating negative journeys or further improving on existing customer experiences. In real time, AI powered analysis offers in the moment guidance to agents actively engaged with customers to improve customer satisfaction and issue resolution, to mitigate regulatory or churn risks, to assist with script and behavioral adherence, and more.”
Kolsky: “Customers interact with organizations in many different ways, and rarely are two interactions the exact same. Using analytics, organizations can spot those changes in real time, and adapt to the customer’s desired interaction instantly. For example, a user of a mobile banking app may use it one day to simply check a balance, but the next day review all recent transactions and maybe dispute a charge.
“So the bank has to look at the two different interactions, how the customer took their journey in two different ways, and then apply analytics to understand the amount of effort it took the deliver both journeys, how many systems were involved, and ways to optimize the interactions. Using analytics, whether the customer goes one way or the other, both directions are optimized and require less time and fewer resources. Analytics is how the company understands the customer journey to deliver an optimized process and have the right resources in place.”
Taylor: “Implementing real-time and predictive analytics brings many benefits to the customer journey. Powerful predictive analytics enable organizations to accurately forecast customers’ needs even, in some cases before those customers know themselves. This empowers contact centers to utilize proactive communications based on these predictions, and to personalize and automate inbound interactions based on customer preferences. The result is that customers are kept happy, thereby minimizing the amount of unplanned inbound contact and improving customer loyalty.
“Predictive analytics can also help organizations with workforce management (WFM), by anticipating when to expect lower or higher levels of customer contact and automatically adjust employee scheduling accordingly. McKinsey estimates that companies have already applied this technology to cut employee costs by up to $5 million per organization.
“Real-time analytics provide live feedback to agents and supervisors on the success of calls. These can incorporate NLP (Natural Language Processing) to parse meaning from speech and use key words to suggest useful information to customer service agents. This type of technology also monitors stress levels using sentiment analysis, and examines script adherence to ensure compliance, all while a call is in progress.”
Herbert: “Real-time predictive analytics is revolutionising customer journeys. By aggregating and analysing customer data, predictive analytics yields valuable predictions about future customer behaviour. These insights enable customer service agents to proactively anticipate customers’ needs, and this ability translates into real benefits for customers, agents and organisations.
“Consider the case of a customer who lacks the technical skills to articulate their issue. The potential for frustration and poor experience is high. Predictive analytics can help an agent discern the customer’s needs and resolve the issue based on similar issues that this customer and other customers have experienced, transforming a difficult situation into a smooth, value-filled customer journey.
“The ability to anticipate customers’ needs transforms journeys into highly contextual, personalised experiences that drive brand loyalty, increase customer satisfaction and achieve rewarding experiences. In a customer journey, predictive analytics lights the path.”
Blanke: “Because the customer journey has historically been pre-defined with a one-size-fits-all approach and with a limited set of inputs, the experience is often inefficient and robotic with customers frequently repeating themselves and leaving the conversation frustrated. Real-time and predictive analytics provide more subtle insights to dynamically influence and intelligently automate and guide the journey – resulting in a more efficient and personalized experience.
“For one, the customer journey lens widens and we pivot to providing more proactive support rather than reactive by looking at real-time data such as web and app activity to determine the ideal time to prompt engagement. Combined with historical and user profile data, we can begin to predict intent to provide self-service capabilities, intelligently route, and dynamically tailor our message with more relevant and personalized content.
“Intent and sentiment monitoring provide live feedback during the conversation to help the agent quickly identify the issue, equip them with answers, and recognize when the conversation is going off-track in order to escalate before issues arise. Real-time and predictive analytics on their own are not enough though, you need automation that is keyed off the data to take action in real-time and orchestrate a more personalized and more contextual customer journey.”
Sherlock: “Today’s customers engage with organisations across a range of communication channels, from social media and email to chat and voice. It’s also no surprise that customers no longer take a linear path to solve their product or service issues. They engage on multiple devices and across channels, while pausing and resuming their journey along the way.
“For businesses, it’s critical to provide smooth, integrated customer experiences across every channel. But understanding customer needs, even as they jump between channels, is easier said than done.
“Predictive analytics enables brands to anticipate the next channel a customer might move to based on past history, making it possible to route customers to the right customer service agent. For example, if a customer often moves from live chat to the phone, organisations can ensure they provide the proper department phone number in the chat when it’s clear the customer is ready to move channels. Further, real-time analytics can help agents improve outcomes during interactions, regardless of channel, to enhance the customer journey, such as identifying cross- or up-sell opportunities while during the conversation.
“Solutions that combine the power of predictive and real-time analytics to make it possible to weave together customer interactions across every touchpoint during a journey.”
Will real-time and predictive analytics end up over-riding live agent empathy or compliment it?
Ziv: “There is something incredibly and inherently valuable about human-to-human interaction. Humans are quite good at being empathetic – and today something that exposes machine limitations and weaknesses. To this end, I don’t believe that we’ll see real-time and predictive analytics trumping live agent empathy anytime soon. But machines can support a key role in enabling our super-human empathy trait to shine through.
“By their very nature, human-to-human interactions are more complex and more emotionally charged. In handling these interactions, agents can be overwhelmed with the myriad tasks they need to perform – showing empathy, understanding sentiment, adhering with compliance requirements, minding average handle time and other key metrics, etc., and in many cases all while juggling work and family life in the home.
“Real-time agent assist solutions can take the onus off agents to make sure they aren’t overlooking anything critical, so they can then focus on meeting the customer’s emotional needs, while all operational and service requirements are taken care of. It should be noted that real-time agent assist can also provide guidance in the moment to other staff to help on difficult calls, such as supervisors or retention specialists in cases of complaints and escalations – this can reduce the stress and strain placed on agents and help achieve better call outcomes.”
“AI chatbots and predictive analytics will begin to “learn” empathetic methods of speaking or chatting by being trained using data from human-to-human interactions, but there can never be a true replacement of live agent empathy. Real-time and predictive analytics complement live agents by guiding them in the moment to help them steer conversations toward positive customer satisfaction, and where empathy is required, a simple prompt can remind them to take a moment to acknowledge the customer’s situation.
Taylor: “As with all forms of Artificial Intelligence (AI) and advanced analytics in the contact center, real-time and predictive analytics should be introduced to complement, rather than replace, human agents.
“Tools such as sentiment analysis can be useful in providing a holistic view of how customers and agents interact and, alongside NLP, offer useful suggestions and relevant information to agents to help them solve queries. But analytics tools cannot replicate the way humans interact with other humans in a completely reliable way.
“Customers are often left feeling disconnected should they feel that their interaction with an organization is dehumanized and robotic. They may choose to move to a provider that offers a more ‘human touch’. As such, your human agents are your most precious tools, and analytics technology should be seen as a way to empower them to deliver a consistently excellent and personalized customer experience, rather than to outshine them.”
Herbert: “Agent empathy, a cornerstone of personalised customer experience (CX), will not be replaced anytime soon. One reason why predictive analytics is so powerful is because of the way it complements agents’ abilities by providing them with an additional toolbox of skills. Predictive analytics enhance agents’ understanding of conversation flow, providing next best actions that enrich interactions.
“With good predictive analytics tools, the agent remains totally focused and attentive to the customer’s needs, thanks to swift access to the right data or process, resulting in fewer and shorter wait times for customers. It can also be used as an indicator to help agents gauge a customer’s emotions, particularly in written form where human instinct may be less attuned.
“There are multiple ways that predictive analytics can work in tandem with skilled agents. It can support an upselling campaign, guiding agents towards the highest likely deal for a given customer. Additionally, it can identify root causes of dissatisfaction and redirect interactions to a supervisor. Agents become more efficient and accurate, and can devote more time to higher-value tasks. Augmented agents receive insights from analytical tools, but the output remains one hundred percent their own. With the aid of predictive analytics, that output gets better.”
Blanke: “More and more evidence point to the fact that the best outcomes are realized when live agents and virtual agents are teamed together for ongoing insights and collaboration. As agents focus on the more complex problem of resolving customer issues or making the sale, they often miss more subtle nuances and clues within the conversation. By equipping Agents with additional insights in advance of and throughout the conversation, they can personalize the interaction and concentrate their focus on solving the problem.”
Sherlock: “The role of live agents has fundamentally changed. While customers often use self-service options for simpler inquiries, voice continues to be the empathy channel, and customers prefer voice interactions for their more complicated or complex issues.
“Analytics and automation capabilities, like AI and ML, have the power to compliment agent empathy in these complex, emotionally-charged interactions.
“For example, real-time analytics can identify when a customer is getting increasingly agitated, based on emotion and acoustical analysis, and give agents tips on how to improve the direction of the call or escalate to a supervisor as needed.
“Given the impacts of the pandemic, this is particularly important. More customers than ever before are facing vulnerability – such as illness, financial hardships, job loss, and more. Analytics can help agents identify vulnerable customers in real-time and offer guidance on how to deal with those customers to be more empathetic and drive better outcomes.”
What could human agents learn from sentiment analysis to benefit agent experience?
Ziv: “Prior to the pandemic, there were ample opportunities for collaboration with contact center peers and supervisors, where agents could listen and learn from others. In the work-from-home environment, agents don’t have that community and support for ongoing learning and coaching. Real-time agent assist solutions can provide an ongoing coaching mechanism and positive reinforcement to support agent engagement at a time when it has never been more needed. It also can help new hires get up to speed quickly and feel more confident in their interactions with customers.”
Einstein: “At NICE, we have come to realize that agents need more than just sentiment analysis – they need AI predictive and interpretive behavior analysis. This is because there is no tangible action to take to improve a sentiment score – so we offer behavior scoring on soft-skill behaviors proven to impact customer satisfaction: Build Rapport, Empathy, Active Listening, Acknowledge Loyalty, Effective Questioning, Demonstrating Ownership, Set Expectations, and Inappropriate Action. NICE Enlighten AI scores each behavior separately, each of which rolls up to the overall Sentiment Score, or Behavior Index.
“This is highly successful because it is now possible for an agent to see which specific behavior may be scoring low and thus drawing down their overall sentiment score, and they can immediately work to self-improve. This also empowers supervisors because they can see the same metrics and can provide personalized coaching. By using always on, always connected data for easy visualization and drill-down, supervisors spend more time on agent empowerment and much less time on random sampling or subjective interpretation.”
Taylor: “Sentiment analysis provides a golden opportunity for contact center leaders to really drill down into their employees’ performance in a way that has not been possible using traditional simplistic KPIs, such as Average Handling Time. These old-school metrics don’t always paint the full picture of agent performance, meaning some employees may not get the recognition they deserve.
“Supervisors can program sentiment analysis to identify those agents who react well when faced with difficult or agitated customers, using intelligent routing capabilities to play on these strengths in the team. In the same way, the technology can help spot ‘subject matter experts’, who are well-versed in a particular area that customers enquire about, and group these agents accordingly. This empowers all agents to focus on customer interactions that best match their skillset.
“At the same time, sentiment analysis can flag stressful interactions, which might have been difficult for an agent to handle, so that supervisors can offer suitable support. For example, Content Guru’s NLP and sentiment analysis technology is already being used by areas of the emergency services to detect where a call agent has dealt with a particularly harrowing interaction.”
Herbert: “Agents need to balance multiple tasks simultaneously, and stressful interactions add another layer of difficulty. It is easy for even skilled agents to overlook some indications as to how conversations are progressing. Real-time sentiment analysis helps detect signals from customers and can provide visual indicators which guide agents on how to proceed. It’s a source of real-time intelligence as to customers’ moods regarding products, services and their overall experience with an organisation.
“In an increasingly connected and multicultural world, there’s also a risk of agents missing cultural nuances that could signal dissatisfaction. Sentiment analysis can help bridge cultural differences by boosting agents’ understanding of customers’ moods. Sentiment analysis plays a similar role in live chats. People sometimes have more trouble discerning emotions in written rather than spoken language, whilst for AI the reverse is true.
“There is one important caveat concerning sentiment analysis: it should be used as an assistance tool, not a monitoring tool – analysing customers, not agents. Supervisors should make sure agents understand that sentiment analysis is not a real-time performance review, rather additional support to better understand customers. Used as such, sentiment analysis can be a powerful aid to talented agents.”
Blanke: “In advance of a conversation, sentiment analysis can inform routing logic to improve Customer and Agent pairing, ensuring the matched Agent is best equipped to manage a particular type of interaction, but it also arms that agent with an understanding of the customer mood to maximize the chance of a successful outcome. Throughout the conversation, sentiment analysis takes the guesswork out of how the conversation is going or how the customer is feeling, providing real-time feedback and guidance to agents. Over time, sentiment analysis can be mapped against customer experience metrics to optimize training and coaching by identifying effective conversation tactics and scripted responses as customer sentiment changes.
Sherlock: “By leveraging AI-fuelled sentiment analysis during an interaction, organisations can offer improvement suggestions to agents for things that they may be unaware of or have simply forgotten in the moment.
“This is powerful because it offers suggestions based on a set of interaction data – far greater than a single person can interpret in a lifetime – while still requiring human agents to use their strengths in critical thinking. This delivers the best of both worlds – well-informed humans focusing on what makes us great, and hyper-focused machines providing insight that we may not have.
“This not only makes outcomes better for customers, but it helps agents become better agents. Through continued engagement with sentiment analysis tools, agents can be more aware of important emotional queues, learn how to better navigate different types of interactions with different customers, and even when to disregard AI suggestions that may not be right.”