Why Accuracy in Natural Language Processing is Crucial to the Future of AI in Retail
More than ever before, artificial intelligence (AI) is being used as a way for people to communicate with brands. Retail companies in particular are turning to AI-powered chatbots as a critical part of their tech stack to deliver mass precision reach and a superior customer experience. In fact, retail is expected to overtake banking to become the industry leader for spending on AI, investing over $3.4 billion this year on a range of use cases, according to IDC. The opportunity for both e-commerce and brick-and-mortar retailers is to leverage these AI interfaces to offer 24/7 customer support, product recommendations, real-time query handling and order assistance.
The first generation of chatbots didn’t have real natural language understanding/processing (NLU/P) capabilities. NLU/P is the deep learning technology that allows chatbots to accurately understand a user, analyze what they say, and generate a response in a human-like way. Most first-generation bots previously couldn’t provide correct responses after a couple of tries, which frustrated early-adopting consumers, businesses and influencers in the space. Consequently, it has become imperative for customer-facing bots to produce near-perfect accuracy. As the omnichannel customer experience becomes more and more of a key brand differentiator, it's crucial for the business to get things right at the first point of contact, helping to drive loyalty and ensuring continued visits.
Only recently have chatbots advanced to the point where they're able to truly and accurately understand natural language text and speech — i.e., understanding the utterances and nuances (and their intentions) that a human would say to another human in their native language. AI-powered chatbots in retail settings must also enable 24/7 customer service responses for routine questions in all major languages so that they can connect with wider, more diverse audiences, and because most major brands have a global footprint.
We all know that customers that have an end-to-end positive experience will not only be loyal to your brand, but they’ll also be willing to promote your brand to friends and family. On the flip side, unhappy customers are also quick to act. A study from Accenture found that half quit doing business with a company immediately after a bad sales/marketing experience, one-quarter took to social media, and 54 percent started engaging with other companies.
Customer support is one of the most important touchpoints for consumers. The most crucial points where chatbots can alleviate any pain points in that process include the following:
- understanding the nuances in a customer request;
- showing empathy for the customer’s problem;
- delivering a personalized conversation in resolving the issue; and
- using it as a opportunity to deliver the brand promise.
None of those functions would be possible without advanced NLU/P technology.
A personalized greeting with their name and asking if they’re calling about an order they recently placed automatically enhances the delight. If the chatbot can anticipate possible issues with the order, it can help navigate the user towards a personalized set of choices or automatically trigger a workflow. A study from Epsilon found that consumers who believe personalized experiences are very appealing are 10 times more likely to be a brand’s most valuable customer — i.e., those expected to make more than 15 transactions in one year. Chatbot accuracy in these situations will help build the customer’s confidence, both in the bot and with the business.
In the omnichannel world, the customer may have placed an order online to pick up in-store, and calls the customer support line for help. To enable the most accurate conversation, it’s imperative for the chatbot to have access to the complete customer context across the different touchpoints. Furthermore, the NLU/P system should not only interpret the user’s intent, but also have the ability to overlay past behavior (context) and identify possible areas of support. On the other hand, if the chatbot cannot correctly identify the intent, it’s important to establish a workflow where the customer can be handed off to a human agent.
Ultimately, it’s the interaction between the customer and the business, and the perception of how the business treats them that leads to the holy grail of customer retention and loyalty. This can only succeed over time with AI-powered chatbots through the use of NLU/P that's put through rigorous field tests. Results from these tests are analyzed to produce a greater level of accuracy each time, and new benchmarks will be established in the industry. This will ensure these intelligent, conversational interfaces are more capable than ever before to identify intent as well as understand and respond to queries just like a human would.
Madhu Mathihalli is the co-founder and chief technology officer of Passage AI, an artificial intelligence company that specializes in conversational interfaces.