How conversational AI is reshaping customer engagement in e-comm - Part 2

In this two-part report, Adgully seeks to explore the nuanced journey of conversational AI’s evolution, examining the milestones and breakthroughs that have shaped its trajectory.

Also read:

Revolutionising customer support - Part 1: The role of conversational AI in e-commerce

Conversational AI has impacted customer engagement and satisfaction in the context of fashion retail. Complex search queries can pose problems when it comes to dealing with customers. Raghu Krishnananda, Chief Product and Technology Officer, Myntra, explains how the fashion retailer dealt with the issue. “When we analysed our search queries on the platform, we saw that about 10% of the queries were complex queries, that is, more than five words. With GenAI, this problem has become much easier to solve and ChatGPT solved this out of the box for us. We leveraged GPT3.5 for query and intent understanding, and coupled that with our own search platform to serve results that include relevant related products to address the queries. We expect an increase in such queries as users get more confidence in the results and become more used to voice search, shopping assistants, etc. in general. Both our GenAI-based solutions, MyFashionGPT and Maya, were built to address this need state. While these are still fairly new, we see good traction in user adoption as well as repeat usage which is a sign that people are finding these features useful to make purchase decisions,” says Krishnananda.

He sees a great future of conversational AI in fashion customer support. Like in many other domains, explains Krishnananda, GenAI is one of the key emerging trends in fashion e-commerce as well. He adds that Myntra sees many opportunities to use this versatile technology to enhance the experience for its customers and brand partners and also improve efficiency.

According to Krishnananda, some potential use cases are:

Use of LLMs for query understanding and customer intent understanding. Expect a transformation from the current shopping bots to styling advisors for beauty and fashion.
Use of GenAI for fashion design to help designers come up with creative designs.
Chat summarisation and sentiment analysis to help CC agents with faster issue resolution.
Automation of creatives for merchandizing and notifications, etc.

The right balance

Achieving the right balance between automation and the human touch in customer support, particularly for fashion-related inquiries, is crucial. Automation can efficiently handle routine tasks and provide quick responses, but it may fall short in addressing complex or personalised fashion queries that require a nuanced understanding of individual preferences and style.

Raghu Krishnananda is certain that the human touch will always be irreplaceable. “We recognise the invaluable role that technology, including AI, plays in enhancing customer experiences and providing efficient support. We also provide chatbot support if customers wish to use that communication channel. At Myntra, we strive to strike a harmonious balance between the capabilities of AI and the genuine warmth and understanding that human interactions bring, ensuring that our customers receive the best of both worlds in their journey with us,” he explains.

Other use cases

Experts highlight the evolution of AI-powered chatbots and virtual assistants, not only streamlining customer service, but also extending to areas like employee training, workforce management, supply chain optimisation, fraud detection, security, and dynamic price optimisation, etc.

In the dynamic landscape of 2024, e-commerce companies face the critical task of ensuring their conversational AI systems meet the highest standards while staying ahead of future trends, states Sheshgiri Kamath, CEO and Co-founder, Kapture CX. This challenge, according to him, encompasses the need for hyper-personalisation in customer support and product recommendations, ensuring that AI-driven interactions are not only efficient, but also deeply attuned to individual customer preferences.

“The focus extends to providing an unparalleled customer experience, enhancing satisfaction, retention, and conversion rates through AI’s nuanced understanding of consumer behaviour. The evolution of AI-powered chatbots and virtual assistants is crucial, streamlining customer service and reducing human agents’ workloads. Beyond customer interactions, AI’s role extends to employee training and workforce management, leveraging analytics for operational efficiency,” Kamath adds.

According to Kamath, some key use cases are:

Key areas like supply chain optimisation: AI will fine-tune demand forecasting, inventory management, and logistics, driving cost reduction and bolstering supply chain efficiency.
Fraud detection and security: AI systems will actively identify and thwart fraudulent activities in both online and in-store transactions.
Price optimisation: Dynamic pricing algorithms will adjust prices in real-time based on demand fluctuations, competitor pricing strategies, and prevailing market conditions.

“By improving demand forecasting and logistics, AI boosts supply chain efficiency and reduces costs. In security, it combats fraudulent activities in online and in-store transactions. Moreover, AI’s dynamic pricing algorithms adjust prices in real-time, reflecting market conditions and competitor strategies. The integration of AI in e-commerce is not just a trend, but a necessity for maintaining a competitive edge. Companies that successfully harness these technologies are likely to emerge as industry leaders. E-commerce companies must not only embrace AI innovations, but also ensure they are implemented responsibly and sustainably, keeping future trends and standards in sharp focus,” Kamath concludes.

Data privacy and security

With data privacy and security being paramount concerns, e-commerce players need to ensure that its conversational AI systems adhere to the highest standards.

In the realm of e-commerce, where data privacy and security are paramount concerns, ensuring that conversational AI systems adhere to the highest standards is crucial for building and maintaining trust with users, says Gaurav Singh, Founder-CEO, Verloop.io. According to him, e-commerce players prioritize robust encryption protocols to safeguard the transmission and storage of sensitive customer data. By implementing end-to-end encryption and secure socket layer (SSL) technologies, they create a secure communication environment, minimising the risk of unauthorized access or data breaches during customer interactions with conversational AI systems.

Furthermore, he adds, comprehensive privacy policies and transparent disclosure practices are integral components of building trust. According to him, e-commerce players invest in clear and easily accessible privacy policies that outline how customer data is collected, used, and protected.

“Communicating these policies in a user-friendly manner ensures that customers are well-informed about the data practices associated with conversational AI interactions. Regular audits and compliance checks, often conducted by dedicated cybersecurity teams, help e-commerce players stay ahead of the evolving privacy regulations, ensuring that their conversational AI systems adhere to the highest standards and align with global data protection laws,” he states.

Additionally, adds Gupta, e-commerce players take proactive measures, such as anonymising and aggregating data when possible, to demonstrate a commitment to privacy. “They work towards minimising the collection of personally identifiable information and emphasising data minimisation principles in the development and deployment of conversational AI systems. By adopting a privacy-by-design approach and continuously enhancing security measures, e-commerce players not only address privacy concerns but also actively contribute to the establishment of a trustworthy and secure environment for their users, reinforcing the integrity of their conversational AI systems,” says Gupta.

In the e-commerce sector, prioritising customer data privacy and security is crucial, says Tarun Dua, CEO, E2E Networks, adding adopting powerful open-source large language models (LLMs) like Llama2, Mistral, or Falcon, and hosting them privately is an effective strategy.

“Cloud GPU platforms facilitate this by allowing e-commerce companies to use these models on their datasets, thus combining high customisation with enhanced data security. This set-up enables chat models to operate directly on company data, improving accuracy and speed in customer interactions. Also, these models can leverage specific data sources and the user’s conversation history for context, leading to informed responses. Building Conversational AI in-house not only enriches the chat experience, but also ensures secure handling of customer data within the company’s infrastructure, thus effectively addressing privacy concerns associated with proprietary AI platforms,” Tarun Dua adds.

Trends to watch out for

Looking ahead, what trends can we foresee in the conversational AI and customer support space?

Gaurav Singh predicts that several trends are poised to shape the conversational AI and customer support landscape.

One notable trend, according to Singh, is the increasing integration of advanced AI technologies, such as natural language processing (NLP), Generative AI and machine learning, to enhance the contextual understanding of user queries. This will result in more sophisticated virtual assistants and chatbots capable of providing nuanced and personalised responses, thereby elevating the overall customer experience.

Additionally, he adds, the rise of omnichannel support is expected to continue, with businesses focusing on seamless transitions between different communication channels, including chat, voice, and social media, to offer customers a consistent and cohesive support experience across various platforms.

“The upcoming stage involves creating not just a conversational AI solution but a comprehensive platform for sales, support, agent tracking, and quality analysis, presenting an exciting prospect. Currently, the industry relies on various platforms for campaign management, customer support, providing agents with necessary documents for query resolution, chat and performance tracking, and conducting quality analysis. Speaking specifically about quality analysis, managers invest significant time in diagnosing whether agents adhere to all compliances. Now, we’re introducing a complete holistic solution that can perform all these tasks swiftly,” concludes Gupta.

E2E Networks’ Dua foresees two major trends:

The rise of Multimodal Generative AI
The integration of Predictive Analytics

“Large Multimodal Models (LMMs) represent a paradigm shift. They allow for interactions that seamlessly combine text and audio-visual prompts, including video. This extends traditional LLM-powered Conversational AI with multi-sensory skills like visual understanding, potentially enabling richer customer interactions. Simultaneously, the open-source ecosystem is exploring the possibilities of combining predictive analytics with LLMs, thus creating the ability to foresee customer needs with incredible accuracy. These systems can not only anticipate, but also understand and contextualise user queries, promising a future of proactive support, where customer issues are resolved before they arise. As Generative AI-powered Conversational AI becomes more mainstream, we will see customer interactions that are rich and highly contextualised to individual customers’ needs,” Dua concludes.