AI in Contact Centers: Bringing Real-Time Intelligence into the Conversation
Creating an engaging and satisfying customer experience can be a real challenge for even the most technologically sophisticated contact center or brand. It all starts with the staff, the training they receive, and crucially the tools that they use. Having tools that deliver a quality voice experience devoid of lag, jitter, and that overall patchiness reminiscent of an early cell phone going through a tunnel. But enriching the overall experience goes much deeper than just audio quality on a call and the savviest of customer experience companies recognize that they need a scope of tools that deliver much more.
Consequently, an entire industry has sprung up around enriching communication services with artificial intelligence. Analysts and early adopters have already taken notice. Accenture for instance has estimated that the introduction of AI will increase business productivity by up to 80% by 2035. Customer service is one of the business functions expected to benefit most from AI, with Pegasystems forecasting a 10-50% reduction in customer churn. AI can work with agents to enrich the customer experience, resolve inquiries faster, and enable better outcomes for brands.
To facilitate these outcomes Nexmo released WebSockets back in 2016 to allow brands to integrate AI engines into their voice experiences. Over the years we’ve seen an array of use cases and opportunities enabled by this functionality, and the available set of capabilities continues to grow.
Real-Time Sentiment Analysis for Deeper Insight
Creating the best possible experience starts with understanding your customer. This can be particularly challenging in a noisy contact center where even the highest level of call quality can be drowned out by the cacophony of agents speaking over one another to be heard by their respective customers. Small nuances or indicators of dissatisfaction can go overlooked. Adding in real-time sentiment analysis reduces the risk of missing something. This feature provides agents and managers with a real-time visual representation of the customer’s experience, indicating how happy or unhappy he is at any given point of the conversation. With this information, agents can more readily recognize when things have taken a negative turn and adjust course, or when things are going well, continue on their current path.
More importantly, thresholds can be set so that an escalation to a manager is automatically triggered when a customer’s satisfaction level dips below a set score. All together these features give agents and managers the tools they need to understand their customer better and subsequently provide the very best experience possible.
Voice Agents to Resolve Inquiries
Though not conceptually new to the market, time to resolution and customers’ expectations around it continue to grow. Customers want to be able to call in, endure a minimal wait time (if at all), and have their inquiries resolved faster than ever before. One way to solve for this is to simply hire more agents, but that option is costly and time consuming. It also runs the risk of hiring to excess simply to address your worst case, most inundated call times. Furthermore, in many instances the customer is looking for a relatively straightforward answer that does not require a lot of conversation, nuance or relationship building.
This is a prime use case for adding virtual agents, bots that address simple FAQs, and routing the more complex questions to skilled human agents. Customers that call in looking for the status of a shipment or billing statement can engage with a virtual agent that gives them the information they need and they never have to wait in queue to speak with a person. Alternatively, if the conversation turns out to be more complex it can be handed off to an agent.
Adding this functionality offers customers a self service option where they can skip the (potential) line and have their inquiry handled right away, creating a better experience. This also frees up resources and allows agents to offer customers more time, discuss complex issues in depth, and focus on building rapport and delivering the very best experience.
Real-Time Translation to Overcome the Language Barrier
When language is the barrier even the most knowledgeable contact center agent is hamstrung from providing the best possible customer service. Traditionally contact centers have addressed this by hiring for language-specific abilities, for example hiring a certain number of agents who are fluent in Spanish or French in anticipation of calls that will require those language skills. The trouble is estimating how many callers will require that language option can be difficult – particularly when you are trying to assess how many language-specific calls you will have at a given time.
Again WebSockets offers a fix, enabling a connection to real-time translation engines so that agents and customers can communicate even as they individually speak in their preferred language. In this case a customer that calls in and speaks (or prefers to speak) only German can have his call handled by an agent who speaks only English. With real-time translation each party will hear the contents of the conversation translated into their preferred language. This can help reduce wait times by expanding available agent pools, helping businesses leverage their existing talent pool to deliver better results and speedier time to resolution for their customers.
Real-Time Transcription for Continual Improvement
Keeping track of an entire conversation when occupied by other tasks is a nearly impossible challenge for even the most skilled multi-tasker. At any given time an agent is expected to find the answer to a customer’s question, keep the customer engaged and happy, input data into whatever system they might be using and of course keep track of what was said during the course of the conversation. Oftentimes this last part is done in an effort to better understand the customer and subsequently build an overall better experience going forward. But with so much going on and so many things to keep track of it is virtually impossible to capture every last detail.
Using real-time transcription overcomes this challenge, allowing businesses to capture the entirety of the conversation and take actions on it, for instance searching for keywords or terms. With this information in hand brands can hone in on what worked and what didn’t. Is there a messaging strategy that needs adjustment? How can we change our call scripts? And so on.
The findings are then folded into training modules ensuring that new agents (and existing ones participating in ongoing training) benefit from this ever expanding knowledge base.
WebSockets Offer More Choice
The beauty of WebSockets is that the use cases listed above are in no way an exhaustive enumeration of what you can do. WebSockets can be added to your voice experience to connect to an AI engine of your choosing. This gives you the flexibility to choose not only the engine that most precisely fits your needs for use cases such as sentiment analysis or transcription but AI that offers entirely different functionality such as autocompletion of forms.
Getting started with building a better customer experience is easy – you can find the documentation to WebSockets here or reach out to us at anytime to learn more.
