Advertisement
05/17/2021

Exclusive Q&A: Having a conversation with customers – via AI

Dan Berthiaume
Senior Editor, Technology
Dan Berthiaume profile picture

The ability to automatically conduct a dialogue with customers is evolving past the level provided by traditional chatbots.

Chain Store Age recently spoke with Dakshi Agrawal, IBM fellow and chief architect for AI, IBM, about how “conversational AI” is changing the way retailers engage customers.

How would you define conversational AI?
Conversational artificial intelligence (AI) is the set of technology, including natural language processing (NLP) and machine learning (ML), that enables computers to understand and respond to a user in natural language. 

Conversational AI technology like NLP is growing rapidly. IBM's just-released Global AI Adoption Index found that almost one in two businesses today are using applications powered by (NLP). Customer service is the top NLP use case, with 52% of global IT professionals reporting that their company is using or considering using NLP solutions to improve customer experience, and it was the use AI case IT professionals were most likely to report that the COVID-19 pandemic has increased their focus on.

Conversational AI requires training data, machine learning, and natural language processing to emulate human interactions, recognize speech and textual inputs, as well as translate the meaning of text across numerous languages.” 

What advantages does conversational AI offer retailers for customer engagement?
Conversational AI helps retailers provide efficient and seamless customer engagement interactions so they can improve their customer relationships and deliver new business value - it's one of the reasons why adoption of technologies like NLP is increasing rapidly. Virtual assistants can minimize wait times and engage with consumers in real-time. 

There are a few top use cases. First, conversational AI can be used to streamline how a retailer empowers its customers to resolve inquiries across channels — including web, messaging and voice. By helping the customer resolve an inquiry on their own, it reduces costs and frees up customer service agents to focus on more complex customer queries. Second, conversational AI can be used to empower customer service agents to respond to questions faster. 

Using AI and natural language processing, agents can search and find key insights from internal knowledge bases and automate workflows via web and mobile chat. This can help agents provide more personalized advice or cross-sell products to increase revenue.

What are some pitfalls of conversational AI deployments to avoid?
First, the complexities of language are one of the biggest challenges in conversational AI. Conversational AI needs to be trained to understand a wide range of diverse languages, dialects and accents, but basic factors like background noises can interfere with AI's ability to understand what a user is saying. 

Emotions, tone, sarcasm and idioms also pose a pain point for virtual assistants or any other conversational AI applications to interpret correctly and respond in tune. Lastly, each business has its own unique phrases or meanings, so an AI model needs to be customized and trained on the specific language of that industry or company. If conversational AI is unable to resolve a particularly complex customer issue, it's important for it to be able to seamlessly pass the customer to a human agent, right in the same channel. It's key to continuously review and optimize conversational AI to ensure you are providing a smooth customer experience.

Conversational AI often relies on large language models that can easily have bias or content issues. Explainability and fairness tools will become expected in NLP products developed for businesses, both off the shelf and custom models, to help mitigate bias and detect drift in the models that are powering conversational AI. 

Security also needs to be top of mind. Since successful conversational AI relies on collecting data to answer user queries, it can be vulnerable to privacy and security breaches. Developing conversational AI apps with high privacy and security standards and monitoring systems will help to build trust among end-users, ultimately increasing virtual assistant usage over time.

How can retailers measure the ROI (quantitative/qualitative) of a conversational AI implementation?
Conversational AI technologies can deliver both top and bottom-line results. A Forrester Consulting study estimated that a large organization implementing virtual agents can achieve $5.50 cost savings per contained conversation. And a recent IBM Institute for Business Value (IBV) study found that every respondent reported that virtual agent technology has contributed to an increase in organization revenue, the average increase being 3%.

Conversational AI can directly impact a retailer's customers, customer service agents, and financial results so it's important to look at the ROI across multiple dimensions.

These include customer satisfaction. There are a variety of reasons why organizations adopt conversational AI, but improving customer experience is by far the most cited so measuring and surveying customers about their satisfaction after interactions not only provides insight into the ROI, but provides feedback on how virtual assistants and voice assistants can be optimized. 

Also, human agent satisfaction, which is when agents who feel valued and empowered with the proper tools and support are more likely to deliver a better experience to customers. Measuring and evaluating the impact of conversational AI on the satisfaction of human agents is key. 

And retailers need to measure financial impact. When you reduce the time it takes customer service agents to resolve contacts, you reduce the cost to serve. Whether a retailer is empowering its agents with conversational AI to reduce their manual effort or to provide customers a virtual assistant for self-service or to conduct the initial intercept, this technology can have a significant financial impact. Important factors to measure include revenue increases, first contact resolution, Net Promoter Score increases, cost per contact and human agent handle time. 

Positive interactions with conversational AI streamline into increased revenue for the organization, while also translating into greater customer satisfaction and loyalty. By augmenting your customer service representatives with the power of conversational AI, consumers can engage faster and more frequently at any time of day with brands, avoiding long wait times and further improving their customer experience; which retailers should see reflected in their increased consumer loyalty and additional revenue from referrals.