Photo: iStock | Userba011d64_201
Do Conversational AI Chatbots Make Better Sales Non-Associates?
Americans think that artificial intelligence-powered conversational chatbots really get them.
New research of 1,000 U.S. consumers by Capterra finds that 67 percent of ChatGPT users feel understood often or always by the AI bot compared to 25 percent of traditional support chatbot users.
Fifty-six percent of the survey’s respondents who have used ChatGPT say they would be likely to shop from a site that offers similar tech. Only 11 percent, however, have used ChatGPT for shopping purposes at this point.
Participants in the study found traditional chatbots wanting.
Fifty-three percent rate their experiences using chatbots as “fair” or “poor.” It’s likely because of these unsatisfying experiences that only 17 percent have used a chatbot to search for products. Just seven percent have used traditional chatbots for product recommendations.
“Most retail chatbots are rule-based bots and are best used for basic functions, like order shipping status or inventory checks,” Molly Burke, senior retail analyst at Capterra, said in a press release. “With natural language processing, better handling of nuance, and a greater ability to personalize responses, conversational AI has the potential to improve chatbot experiences by simulating the personalization and creativity provided by human agents.”
The ability to communicate with the technology makes the prospects of conversational chatbots exciting to many. Online shoppers are accustomed to AI tracking their browsing and shopping behavior in the background to make product recommendations intended to improve the customer experience.
Retailers, brands and shopping platforms are pursuing the opportunities that ChatGPT and similar apps represent.
Instacart is among those working with OpenAI’s API service to develop use cases for ChatGPT on its platform.
The retail delivery service plans to roll out a new search engine using ChatGPT to answer user questions about food ingredients, healthy meal options and recipe ideas. Responses to queries will come in a dialogue rather than a list of results used in traditional searches. The new “Ask Instacart” service is slated to roll out later this year.
“When you think about grocery shopping, it takes a lot of thinking and planning,” JJ Zhuang, Instacart’s chief architect, told The Wall Street Journal. “It’s the perfect use case for smart AI because it’s a lot of cognitive load.”
Discussion Questions
DISCUSSION QUESTIONS: How much will conversational chatbots influence where consumers do their online shopping searches? How much will it influence purchasing behavior?
Poll
Loading …BrainTrust
Mark Ryski
Founder, CEO & Author, HeadCount Corporation
DeAnn Campbell
Head of Retail Insights, AAG Consulting Group
On their current meteoric trajectory, conversational chatbots will greatly influence online shopping and purchase behaviors in the future. Ultimately, shoppers want timely, accurate information. If this information can be delivered more effectively by chatbots, that can’t help but have an impact. And while there will continue to be growing pains as this technology matures, it’s getting better by the day.
If there is an interface that allows retailers to enter product literature into ChatGPT they could potentially “train” it to be a reasonably good sales non-associate. As it is today, ChatGPT is reasonably good at conversational interactions, but it still needs curation from a knowledgeable human. That said, many retailers’ sales associates come up short as well. For many shopping experiences that is fine, but nothing beats working with a knowledgeable and professional sales associate.
I find too many pre-programmed chatbots programmed to be sure a customer never talks to a human. Am I alone? These pre-programmed chatbots always seem to want to solve the problem I don’t have.
I look forward to AI generated responses. Maybe they will actually listen to me. How often do they ask you how they can help you? Then you tell them, and they give you three totally unrelated options. UGH!
Consumers don’t actively seek out chatbots when they are shopping, so any influence that AI based searches or chats have will be secondary, not driven by consumer choice. Shoppers decide where and how to shop because of many factors: mood, convenience, time, being bored at work, wanting to get out of the house for a bit. Technology plays very little role in the choice. You don’t shop because you feel like talking to a chatbot, you shop because you want to browse products, stock up on groceries, hire a service, or find a new shirt for a special occasion. Smart brands and retailers will leverage AI as part of giving shoppers the best possible engagement experience, but when executed properly AI will be largely invisible to the end user because it delivers an easy and effortless experience.
The problem with applying generative AI-powered chatbots is the lack of situational prompts. From my perspective, two challenges have to be solved before it goes to mass adoption:
1. Shoppers behavior modeling to make communication relevant for every shopper and shopping case;
2. Retailer’s context(knowledge base) turned into a generative AI prompt to make responses sharpened to retailer specifics and specific shopper requests.
I have been thoroughly impressed by ChatGPT’s ability to be informative and helpful over a wide range of frankly complicated topics (ranging from tech, health care and legal). Still, I look at all answers it delivers skeptically knowing that how I asked the questions highly influences the usefulness of the answer. For bread-and-butter simple questions, ChatGPT is probably already ahead of a real (and largely nonexistent) sales associate, but answering those sorts of questions probably didn’t need generative AI’s capability in the first place. How many “daily shoppers” will apply critical thinking to the more complex issues where generative AI shines?
As expressed by Instacart, ChatGPT will be used to do what it does best, search through volumes of data extremely quickly and return the response in a dialogue instead of a results list. This is an improvement on existing search features but far from preventing the frustration many people feel when engaging with chatbots and hoping for a human interaction. Given that product detail pages already have vast amounts of information and that people generally conduct research on their own prior to purchase especially on higher ticket items, I’m not sure how much it will influence purchasing behavior. However this is just the beginning and as complementary technologies continue to evolve, the interactions with chatbots or digital sales people will become incredibly human (and probably as frustrating as they can be with real, in-store humans).
This is one place where AI is going to be transformational. My litmus test for shopping technology is, does the tech reduce friction and improve customer satisfaction? Chatbots and AI absolutely have the capability to do just that. It should give consumers faster results, especially when they’re shopping for items, like in the grocery list example, that may go together or that in the past required multiple searches to get a complete list.
I keep reading about situations where AI chatbots are taught to act like humans. I wonder if when I get frustrated with an AI chatbot it will yell back at me.
The new conversational AI should improve the experience and it has tremendous upside, especially with its capability to search through massive amounts of data and interpret logical results with speed. I’m sure many of you are like me, where I’d rate most of my interactions with non-generative AI chatbots to be poor to extremely disappointing. The future is bright and exciting!
All of the chatbots I have encountered online have been very rudimentary and I try to avoid them. I am sure they will get better, and sooner or later will add value to the shopping experience–but right now I think the only thing they contribute to is higher profit margins, but no topline revenue growth based on the “good service” they provide.
Open AI and ChatGPT are going to significantly impact the self-service e-commerce world. When customers can talk conversationally with a bot, it changes the game. This new generation of AI can make better (personalized) suggestions based on the conversation. I’m personally excited to see how this positively impacts the customer experience. This is almost as big as the invention of the Internet!
Let’s hope the hype is real. I’d love to stop yelling “Operator! Operator!” to escape chatbot jail.
At the end of the day it is how good the chatbots are in conveying accurate information. It isn’t about pretending to be human in the delivery of the information (you shouldn’t expect simulated empathy), but if conversational technology helps understand the questions from humans better, it goes a long way
Fifty-six percent of the survey’s respondents who have used ChatGPT say they would be likely to shop from a site that offers similar tech. Only 11 percent, however, have used ChatGPT for shopping purposes at this point. (emphasis added)
Is today the day to complain about “theoretical” studies ?? I’ll be more impressed when the latter number actually catches up with the first.
Today’s retail Chatbots are mind-numbingly frustrating. To the point where the customer relationship to the brand experience and reputation is being badly damaged.
Most retail companies should yank back control, shut off the bots, and give it to their customer service team until ChatGPT can be properly integrated within their four walls for their product, sales and task-oriented categories.
Once uniquely set up to a retailer, ChatGPT will help change and enhance customer service, will influence shopping habits and increase loyalty.
For me, time will have to tell on this one. If you look at the history of call centres as an example, many moved off-shore from English speaking countries (UK & US) to India starting around 20 years ago. This was very much cost-driven and after a number of years some organisations realised that customers sometimes struggled with accents and poor line quality and began re-shoring these.
However, this is not always the answer. This morning when attempting to address a problem with my bank I texted with a chatbot before being passed to a human who eventually told me to phone a helpline. The helpline told me there was a 24 minute queue – after which I spoke to an agent – my query was resolved in less than 90 seconds. In total the time I invested was over an hour – where as the bank invested 90 seconds. Customer service engagement still has a long way to go – if chatbots are to augment or replace human sellers without frustrating customers then huge advances need to be made.