At Your Service: The Telco Chatbots

Subscribe To Download This Insight

3Q 2017 | IN-4710

The Chatbot that started as a simple lab invention has today turned into a quintessential business asset. Eliza was the first chatbot rolled out of MIT labs by Prof. Joseph Weizenbaum in 1966 as a simple program that used a “pattern matching” algorithm to imitate humans – like communication of a psychotherapist to a patient. The invention, although regarded a breakthrough, did not gain sufficient traction back then. However, the advent of ubiquitous computers, smartphones, gadgets, social apps, and advancements in AI technology in past two decades has completely turned around the scope and scale of chatbots. Apple’s Siri, Microsoft’s Cortana, IBM’s Watson, Google’s Now and Amazon’s Alexa have transformed our conversational preferences from personal to virtual. Investment in AI chatbots is projected to reach more than USD$20 billion by 2021 spanning from retail, healthcare, hospitality to banking, insurance and telco industries.

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.

 

Journey from Eliza to Alexa

NEWS


The Chatbot that started as a simple lab invention has today turned into a quintessential business asset. Eliza was the first chatbot rolled out of MIT labs by Prof. Joseph Weizenbaum in 1966 as a simple program that used a “pattern matching” algorithm to imitate humans – like communication of a psychotherapist to a patient. The invention, although regarded a breakthrough, did not gain sufficient traction back then. However, the advent of ubiquitous computers, smartphones, gadgets, social apps, and advancements in AI technology in past two decades has completely turned around the scope and scale of chatbots. Apple’s Siri, Microsoft’s Cortana, IBM’s Watson, Google’s Now and Amazon’s Alexa have transformed our conversational preferences from personal to virtual. Investment in AI chatbots is projected to reach more than USD$20 billion by 2021 spanning from retail, healthcare, hospitality to banking, insurance and telco industries.

Today’s chatbots are conversational interfaces that are powered by AI approaches like natural language processing (NLP), machine learning, image, video, and voice processing to extract contextual meaning and user intent, and provide human-like, real-time intelligent responses. The end goal of chatbots has evolved into developing cognitive computing abilities from conversational data to analyze human emotions and provide enhanced personal experience during interactions, to self-learn and become a predictor of human preferences. 

Are Telcos and Chatbots a perfect match for each other?

IMPACT


The success of any AI technology is primarily based on the magnitude of data it analyzes. Chatbots in particular run on unsupervised algorithms like NLP and other machine learning techniques and require copious amounts of data to successfully analyze, learn, and perform predictive and prescriptive tasks in real-time. The telcos are sitting on a massive pyramid of diversified data that is accumulated across its business, customer, and network levels both real-time and over time. But a more compelling business case for telcos to integrate chatbots is that its business peripheral stems from its consumer’s behavior. Historically, telcos’ strategies had always been net-centric. However, the influx of smart devices combined with OTT applications has completely changed the demands and preferences of telco customers in last two decades. They consider telcos as an intermediary that facilitates their ever-increasing demand of uninterrupted connectivity for OTT services (e.g. social and media platforms). Consequently, their loyalty to telcos has also become fairly elastic. Hence, telcos are forced to realign strategies to become more customer-centric and compete on dimensions such as customer experience, service assurance, loyalty, and customer retention.

This is where chatbots aim to match their expectations.  Chatbots can holistically connect with customers through omnichannel interactive engagements (text, website, voice, video). They can be programmed to understand customer preferences, usage patterns, analyze sentiments and concerns through contextual interactions and map with relevant products for sales and marketing or troubleshoot network, payments, or service issues on the fly. For more complex problems, they can aggregate statistics and hand-off conversation to a human for further intervention.

… and happily, ever after?

COMMENTARY


Indeed, most of the tier 1 telcos have chatbots embedded within their business support system (BSS). Telcos find it relatively easy to deploy chatbots within a BSS framework since the technology is highly customizable, has matured presence in the market across other industries, and there are many start-ups as well as legacy vendors offering chatbots. For example,Orange’s Djingo provides full ecosystem of content and services using voice or text and also manages connected devices. NTT’s Cotcha provides business solutions for labor shortages and diversification of contact points. Recently released Vodafone’s TOBi (powered by IBM Watson and LivePerson technologies), is a one-stop shop for all service and subscription related queries. Value proposition is justified with reduced operating expenses, increased customer retention, and expanding revenue. Telcos view chatbot implementation as one of their leaps towards digital transformation.

However, chatbots are not free from problems. AI technology that elevates chatbots as versatile customer assistants is also riddled with challenges of its own. Chatbots in service domains apply retrieval based models wherein feedbacks are generated in advance or in accordance with certain patterns. This feedback is based on accurately determining user intent. If a chatbot is not able to correlate contextual information and co-references during interactions, it could give incorrect responses that could potentially frustrate customers. Similarly, if the AI algorithms do not accurately and sufficiently map context to content, then it could confuse customers with incorrect information. Hence, accuracy of NLP models is of primary importance, and telcos should focus on this by rigorously testing validity and predictive capabilities of chatbots as they incorporate them into customer service operations. After all, the cost to pilot and train chatbots in various use cases is relatively low compared to other AI initiatives, and the potential benefits are significant.

Services

Companies Mentioned