Why AI Needs a Human Voice to Realize Productivity Goals


A picture of a drowning person being saved at sea

Rafael Cortes of Foehn investigates how AI is enhancing the role of the agent and human voice and how that goes against previous predictions.

Some years ago, when AI was making its early impact on contact centre operations, stories concerning man versus machine proliferated everywhere. Predictions of highly automated, agent-free contact centres appeared to overwhelm the counter-argument that favoured voice and the human skills of the agent.

For example, two years ago, Gartner predicted that, by 2020, nearly 85% of customer interactions will not be performed by humans, resulting in dramatically improved contact centre performance. From our experience, dealing with business at all levels, both parts of that prediction now look premature.

Furthermore, numerous surveys over the past twelve months still show that customers continue to rate a voice conversation with an agent as their preferred channel of communication.

For example, recent research from Genesys estimates that 75% of consumers still believe humans provide the most effective customer service. That’s why our development team has focused on optimizing voice, as a priority over automation and channel options, in the design of our cloud contact centre system.

Of course, voice communication can be delivered by machines and software as well as humans. Natural voice recognition, intelligent IVR, proactive voice messaging, AI and digital personal assistants are blurring the boundaries of human and digital voice.

Uniquely, though, voice managed by a skilled agent has the power to convey tone, warmth and likability in a way that can emphasize emotional brand values and enhance the customer experience far beyond the capabilities of digital alternatives. Where conversations are sensitive, urgent, emotional or commercially valuable, voice will always play a dominant role.

In practice, though, the ‘AI versus Voice’ debate appears to be moving from conflict to mediation. These two fundamental drivers of customer experience are finding a middle ground where, working together in harmony, they are taking contact centre performance to exciting and unprecedented heights.

Specifically, speech analytics and chatbots are two areas where the integration of AI and voice is giving the agent new powers.

With these technologies now maturing, it’s worth an update on how they can help the agent in the pressured environment of a typical contact centre.

Voice and AI in Speech Analytics

With the development of natural language processing (NLP) technologies and the ability to capture and analyse all kinds of unstructured data, such as call recordings, (in addition to information from traditional structured data sources), businesses are recognizing that there’s significant value in the mountains of ‘voice data’ they collect.

The increasing sophistication of NLP algorithms means that call centre systems can now interpret long strings of words rather than single key words, and route callers to the correct department. This means analytics can understand how sentences are spoken and their underlying meaning.

The software can analyse the caller’s tone, vocabulary and mood to establish emotion and satisfaction levels. It can even detect a caller’s age, which can help to ensure the call is routed to the right type of agent.

When combined with predictive analytics, systems can also identify when a caller is getting frustrated or angry, or when they are lying, and even when they are trying to commit fraud. These insights help agents to handle calls in the best way and improve the overall customer experience.

AI is powering speech analytics to provide ground-breaking insights and predictive observations. This new-found intelligence gives agents the capability to plan and anticipate interactions with significantly higher probability of a positive outcome.

The vast amount of voice stored in the contact centre offers a ‘goldmine’ of data that can feed AI-driven analytics and expose customer behaviour.

For example:

  • Improve first call resolution

As a call progresses, AI can predict the outcome and direction the interaction will take. With this insight, the agent can take action to terminate or prolong the call to make best use of time.

  • Identify CX drivers

By analysing spoken dialogue and acoustic measurements, AI can identify the most powerful drivers of customer experience. This data can be used tactically support points on the customer journey both inside and outside the contact centre.

  • Raise customer satisfaction

AI has the capability to predict customer service metrics, such as Net Promoter Score, C-Sat Scores and Customer Effort Score. Armed with this information, agents are forewarned of the size of the challenge to build customer satisfaction for any given customer.

  • Lower customer churn

Using historical data, AI can identify customers at risk of leaving and can indicate the proactive action necessary to reduce churn.

  • Spot up-sell opportunities

AI can identify the characteristics of potential up-sell customers and inform agents how best to close a sale for a given customer persona.

  • Train chatbots

Speech transcriptions and voice data offer excellent material for ‘training’ the machine learning process of the chatbot.

  • Prevent fraud

AI can detect fraudulent activity based on the questions a person asks and the specific words, phrases and persuasive techniques used.

Voice vs Chatbots

Widescale adoption of chatbots, powered by machine learning, has been driven by the belief that they provide a cheaper, quicker customer experience compared with voice.

On the contrary, voice has key advantages, for example:

  • Calls win on cost

Agents can manage two live chats but with three, the time spent switching is wasted, productivity plummets and cost increases. Typically, calls are half the length of a live chat session and, correspondingly, cost less.

  • Calls are more caring

A voice call can build an emotional connection. Chat can’t. Whereas a phone call is one-to-one, customers know that chat is one to many, leaving them feeling less special.

  • Chat can lose focus

Customers on chat often think agents are available indefinitely and do not hesitate to make a coffee mid-session. The extended disjointed session can greatly reduce agent productivity.

  • Chat lacks ownership

An agent is fully focused on a phone call but becomes less conscientious when handling multiple concurrent chats. This can impact agent motivation and limit job satisfaction.

  • Bots aren’t bad

A survey by Genesys has established that voice is the preferred channel of communication, but over 85% of people will use a chatbot, too. The challenge is to have chat available when it suits.

To Sum Up

The synergy between human voice and AI-powered processes is evolving into one of the most exciting partnerships in the contact centre.

With the ‘best of both worlds’, agents can deliver the essential human interactions demanded by customers whilst also fulfilling the productivity goals of the business.

It starts with ‘getting voice right’, then building the layer of intelligent, time-saving capabilities of AI on top. After that, the challenge becomes one of balance, ensuring that the human interactions with customers always take priority.

This is Foehn’s philosophy and, if you are considering upgrading your contact centre capabilities, you can find some useful points here in our latest guide on Cloud Contact Centre Success Strategies.

To find out more about Foehn, visit: www.foehn.co.uk

Published On: 25th Jul 2019 - Last modified: 18th Sep 2019
Read more about - Industry Insights,


Get the latest exciting call centre reports, specialist whitepapers, interesting case-studies and industry events straight to your inbox.