How can the use of conversational AI help grow your ecommerce business? We spoke with Bharat Gupta, an expert in the field with years of experience fine-tuning and implementing chatbots, to learn about the impact this valuable tech is having on the ecommerce industry, as well as what the future holds as conversational AI continues to advance.
Getting to the heart of AI means moving past the buzzwords and taking a look at how this tech is being put to use by ecommerce businesses right now. In recent years, chatbots have evolved from just answering direct questions from a list of options. Now, they actually behave more like humans, so far as to understand the user’s underlying intent. This makes it an almost indispensable tool for growing ecommerce businesses who need to interact with large numbers of customers at multiple points during a sale, 24 hours a day.
Appeal of conversational AI
The new age of chatbots are able to bring a lot of empathy into the language they use with customers, which can have a direct and significant impact on customer LTV. What conversational AI does really well is present a very interactive, customized experience to the user. This demonstrates a real interest in the customer experience, whether it’s before, during, or after the sale. For small businesses, this can still be managed by a human, but when a business starts to scale, conversational AI can magnify that positive impact to the bottom line while retaining the sense of a human touch.
Applying conversational AI to ecommerce
Considering use cases help us see how this technology can benefit ecommerce business. To get a closer look at this process, Bharat shared three prominent use cases specific to ecommerce sales.
When customers walked into a brick and mortar store, they could usually expect to get advice on products through a tailored interaction with an employee. Even though customers are getting increasingly comfortable buying online, and often prefer to, many report that to an extent they miss that human interaction where they could form conversational relationships. These chatbots act like a salesperson, furnishing curated recommendations based on specific user needs.
To put yourself in the customer’s shoes, imagine you need to shop for an item on a global ecommerce marketplace like Amazon. You enter your search and are bombarded with thousands of results and nearly as many options to filter them by. Rather than putting the cognitive load directly on you, the user, this virtual assistant will gather information on your behavior and preferences through a conversational approach to help curate your shopping experience.
In terms of conversions for an ecommerce store, one global survey revealed that if a user has had a positive shopping experience through a conversational AI, they are 25% more likely to patronize that business.
Collecting 3rd party data
Another way conversational AI can help ecommerce businesses is generating highly qualified leads by collecting third party data. This helps the business to personalize the whole experience of users and not just when the user is interacting with the product or website. This helps with things like sending personalized nudges or advertising campaign emails to customers in order to remarket to them.
Third party data is information that users voluntarily and deliberately share with the business. It features data points such as their favorite sports team, favorite cuisines, and whether they have a pet and what kind. All this is generally acquired through polls and quizzes presented through a conversational interaction.
This data can be used to build personalized and specific product suggestions as well as targeted marketing for each customer at scale, contributing to customer retention and resulting in a much higher LTV.
As acquiring customers continues to get more expensive, the focus for many businesses is shifting to customer retention and improving LTV.Conversational AI can help ecommerce businesses with this by providing 24/7 post-purchase support availability that enhances the customer experience and boosts loyalty. In fact, the convenience, lack of wait time, and non-invasive conversational methods provided by chatbots have been a driving factor in a 24% improvement in customer satisfaction scores.
The big 4
There are four major players who provide Natural Language Processing (NLP): Amazon Lex, Google Dialogflow, IBM Watson, and Microsoft Azure. The exponential rise in text messaging capabilities over the last two years, as well as advancements in contextual understanding, dialogue management, and natural language generation have all contributed to the development of these frameworks. The popularity of the big four allows developers to build for natural language and robust customizations.
The good news is that you don’t really need to have any coding skills to implement chatbots as they are very simple to make. In the end, which framework you choose to use is largely up to personal preference as the conversational capabilities don’t differ greatly from one framework to the next.
Effect on sales
A Forrester survey showed that ecommerce businesses in 2021 saw an average increase in sales of about 67% from users who interacted with chatbots at some point during the sale. About 26% of those sales were started with a chatbot interaction. Noticing this growth, brands like Bharat’s began to experiment with the conversation-first approach to chatbots. But it wasn’t an instantaneous process.
Within the last six years, nearly 87% of surveyed ecommerce customers who interacted with chatbots claimed to have had frustrating experiences. Chatbots first started off as simple, rule-based bots. However, thanks to companies like Amazon and Google making their bot frameworks open source, conversational AI has become much more intelligent. By 2021 the number of frustrated users came down to about 65%, but that’s still quite a large number. So why such gradual progress?
Implementing conversational AI
Bharat and his company ended up using Google’s Dialogflow to create their chatbot. One massive advantage of Dialogflow is that it leverages billions of user interactions from Google Assistant on Android devices to power the NLP capabilities of its AI. Another is that it provides a web interface to create bots, making it easy even for non-techies to create basic bots. Actions are easy to understand and configure, allowing you to make a basic Q&A chatbot within a few hours.
Things get more complicated when you get to specific use cases, which require more training. Bharat found that every chat bot had to be trained separately, and ended up using multiple open source automation scripts to help with the process. Initially, it took him and his team about two to three months in order to achieve the industry standard 90% relevancy of chatbot responses.
Another metric for gauging the performance of chatbots is the percentage of interactions that default to a human interaction. These are instances where the capabilities of the conversational AI were not enough and the user demanded a human interaction. It’s expected that some human interaction will be needed for certain cases, but getting that number down will save money and make the most use of the AI. This fine-tuning is an ongoing process, and it took Bharat’s AI roughly one and a half years to achieve handling 52% of interactions without defaulting.
Tracking these two metrics as well as the average interactions per chatbot session are good KPIs for monitoring your AI. Bharat suggests checking these metrics regularly, at the very least monthly, to make sure the AI’s capabilities are improving. Overall, have patience, and remember that it’s a journey, not a destination, as bots continue to improve with their adaptive learning.
Future of conversational AI
Nearly 1.4 billion people have interacted with chatbots which amounts to over 20 billion chatbot interactions overall. With businesses projected to spend about $78 billion on cognitive AI systems and the rise of Web3, conversational AI can play a big role as blockchain technology and virtual reality continue to develop. And while chatbot interactions can differ from chat to voice to video, the conversational design and algorithm structures remain the same. So as Bharat said, have patience and faith in your AI, and enjoy the journey as your AI learns and grows with your business.