Nuances of Voice Technology and its Value to Businesses and End Customers
By Sourabh Gupta, CEO and CO-Founder of Skit
Modern organizations are constantly on the lookout for new ways to enhance customer experience without escalating costs. Among many different options, voice technology has gained immense popularity over the last few years, across customer communication and lead generation. By 2025, it won’t be surprising when one finds themselves greeted by an intelligent voice assistant or voicebot on their bank or insurer’s helpline. And the difference between conversing with a bot and a human agent will be negligible. A voicebot can ask for inputs based on caller query in a free-flowing manner, replicating a natural, human-like conversation in the customer’s preferred language. This is only one of the several ways Voice AI can be leveraged by businesses. To think that Voice AI’s only application is limited to replacing IVRs would narrow the potential of this amazing technology.
Before we delve further, a clarification. The kind of voice technology that is being discussed in this article refers not to consumer voice assistants—such as Siri, Alexa and Google Assistant—but to Voice AI that helps businesses with a cost-effective approach to connecting with a larger customer-base and offering flawless experiences.
State of Sophistication today
Voice AI has reached a level of sophistication that was unimaginable two or three years ago. In a nutshell, Voice AI today is able to hold convincing human-like conversations. The technology is able to:
- * Hyper-personalize every conversation
- * Identify speaker context and intent
- * Understand different languages and dialects
- * Gauge sentiments, intent and context from the way the other person is speaking
- * Switch language and dialect based on the other person’s preference and comfort
- * Pause to listen when the other person starts speaking
- * Pick up the conversation exactly where it was left previously, including abrupt call disconnects
Leading AI/ML companies are striving to optimize for this level of expertise so that engagements with a customer can achieve truly heightened satisfaction scores.
The Potential of Voice AI in Augmenting CX Efforts
| Approach | Applications | Outcome | What is in it for Business | What is in it for the end customers |
| Inbound | Making complicated IVRs customer friendly with a voicebot, for customer queries | – Reduction of average call handling time (AHT) by upto 30%, by connecting directly to the correct IVR node
– Zero to minimum leakages – Automate mundane, repeated queries handled by agents |
– Direct cost saving to the tune of 60%
– Improved customer experience – Agents can be assigned to perform more complex tasks, leading to higher employee satisfaction, low attrition, lower training costs |
– Enhanced CX
– Minimum possible time used to reach resolution – Less customer frustration in finding correct node/solution – Self-serve options, available at their time of convenience |
| Outbound (Marketing) | – Personalized promotions – Lead qualification calls- Subscription renewal reminders, persistency – Upsell/ cross-sell calls |
– On demand, scalable, and reliable | Capability to reach out to millions of potential customers in a cost effective way | – Can choose to connect at their time of convenience
– Can get detailed information on-demand |
| Outbound (Customer engagement and servicing) | – Validation calls
– Welcome calls – Engagement calls – Greeting calls for occasions – Service disruption and/or fraud communications – Service calls |
Proactive communication leads to increase in customer loyalty and trust | Capability to reach out to millions of customers in a cost effective way in order to make them feel cared and engaged to increase customer loyalty. | Feeling of getting cared by their service provider |
An Informed Approach to Implementing Voice AI
For companies that are deploying Voice AI for the first time, it may seem very easy to build a solution by connecting various parts of the technology. For example, there are several open source Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) engines that can be connected to build voice AI in a matter of a few months. While this is a good approach for developing a generic solution, our experience across extensive implementation and groundwork on deploying voice tech in various organizations points to the fact that the whole is more complicated than the sum of its parts. There are multiple layers that will need to be considered while building a business-specific solution. And it’s all about nuance.
- * For a country as diverse as India, one needs to understand that the English language spoken in India is fundamentally very different from the way it is spoken anywhere else in the world. Grammar, accents, disfluencies, language mix (Hinglish, for example), and organic language switching, all play a crucial role in enhancing the accuracy of a voicebot, when location or demographics influence the use case.
- * The voice tech deployed should be able to support not only the languages that customers predominantly speak, but also specific dialects, for a more ‘comprehensive’ experience.
- * Typically, third-party voice tech vendors spend considerable time in understanding a company, its processes, and the industry landscape, in order to set up, train, integrate, and deploy the bot. While implementing plug-and-play applications can help deploy pilot programmes in the shortest possible timelines, a lot of implementations fail in this approach as there is a low level of customization offered, and high amounts of uncertainty and risk involved.
- * Domain-expertise is another important aspect to consider. Generic speech recognition technologies are extremely poor at detecting intents that are very specific to certain domains. It requires a lot of experience and effort to train the bot to achieve acceptable levels of accuracy.
- * Voice AI needs to be designed to integrate with various systems in real time, to achieve the level of personalization needed for conversations with each and every customer. The underlying stack should also be able to analyze customer speech patterns and derive useful business insights like the customer’s age, gender, region, dialect, tonality, preference for focused targeting and enhanced experiences
- * The ideal Voice AI solution should be enterprise-ready, especially in terms of agile development processes, and robust security and compliance-handling.
Organizations that do not internalize these nuances will end up with solutions that may have poor accuracy and performance levels. This would eventually result in failed voice automation projects, adversely impacting customer experience and overall operational efficiency. A key area that could potentially drive significant business value for organizations globally is Conversational AI. Mastering Conversational AI will deliver more contextual dialogue between people and brands, and as a result, build a more relevant and intimate relationship between the two.
