Conversational commerce has emerged as a groundbreaking trend in ecommerce, marking a shift towards increasingly interactive and intuitive shopping experiences. This movement towards conversation-driven transactions is fueled by both the modern shopper’s expectation to be understood and catered to in a personalized manner and by recent advances in GenAI.
In fact, according to a recent Forrester survey among US retail and wholesale digital experience professionals, 86% are using or experimenting with GenAI to enhance or transform their customers’ digital experience.
Does that mean in the future most commerce websites will expect shoppers to interact with a chatbot?
Actually, the popular view that conversational commerce is synonymous with chatbot-based interfaces is a myth that needs to be debunked. While chatbots represent a way to introduce more dynamic interactions, they are not the only or most effective means – as many shoppers will agree – to achieve true conversational commerce.
This blog post explores the risks resulting from integrating chatbots in the context of product discovery. We highlight the potential pitfalls and propose an alternative approach that leverages the robustness and effectiveness of traditional search interfaces while embracing some of the unique advantages of conversational commerce.
Five Common Pitfalls of Chatbots in Ecommerce
1. Disruption of Customer Experience
Research from Forrester reveals that 90% of shoppers use search functions on ecommerce websites, underscoring the critical role of traditional search interfaces in the online shopping experience. With their clear ranking of results, these interfaces have stood the test of time. Consider that visitors using the search function convert at 4.63% — more than double than the average of 2%.
Known for their sometimes less-than-satisfactory interactions, introducing chatbots risks disrupting your customers’ experience due to an unfamiliar interface and additional friction, potentially leading to decreased conversion rates and engagement.
One critical issue that also arises with the integration of chatbots in ecommerce is the potential for speed losses during information retrieval. In the fast-paced world of online shopping, shoppers have come to expect nearly instantaneous access to information. Speed is crucial for a seamless customer experience. Chatbots, despite offering a conversational interface that might enrich the shopping experience for complex inquiries, tend to lag in response times, which is especially annoying for more straightforward queries.
For instance, a shopper in search of a specific item like a “Stutterheim jacket green” expects quick, direct results. With a chatbot, however, the same query might result in a noticeable delay as the bot processes the request, often generating lengthy and sometimes unnecessary responses about the product. Even if it’s just a few seconds longer than what a traditional search would take, this delay can frustrate users accustomed to the immediacy provided by conventional search engines.
2. Loss of Depth in Product Discovery
Chat-based shopping assistants may look enticing, but they risk simplifying the rich, multifaceted search experience of top-tier websites, which thrive on helpful, rich filters and user control. Traditional search interfaces allow for a nuanced exploration of products and complex catalogs through facets like price, ratings, and categories, enabling both precision and serendipitous discovery. In contrast, chatbots often offer a single product and diminish the user’s active engagement and control over their search. Balancing the ability to handle conversations with the depth of traditional search is crucial to preserving the richness of the online shopping experience, ensuring it remains a helpful journey of discovery.
3. Redundancy and Reinvention
Enhancing chatbots to offer more complex functionalities essentially leads to the recreation of traditional search interfaces. Imagine a future where chatbots don’t simply provide a singular product or a few products but rather present users with a list of options, perhaps even meticulously ranked by their relevance and attractiveness. This advanced chatbot could not only suggest a number of products but also introduce the ability to apply filters and offer refined queries to aid in the user’s discovery process, mirroring the depth and versatility of traditional search interfaces.
While the idea of such sophisticated chatbots is interesting, it raises a fundamental question: Are we merely circling back to the traditional search interface’s proven model under the promise of innovation? By incrementally injecting into chatbots the capabilities that are hallmarks of search engines – such as nuanced query handling, dynamic filtering, and the presentation of a multi-faceted results page – we essentially replicate the very system we aimed to transcend. This project, while ambitious, inadvertently underscores the inherent efficacy and familiarity of the search interface as an indispensable tool in the ecommerce landscape.
4. Historical Precedents
Remember the hype surrounding voice assistants? There were bold claims that voice search would constitute 50% of all searches, heralding a significant shift in digital interactions. However, this transformation didn’t materialize as expected. The limitations of voice search, particularly in complex scenarios in ecommerce that demand filtering, and comparative analysis, highlights the indispensable nature of traditional search interfaces. These interfaces offer a rich experience that voice search couldn’t replicate, especially in displaying multiple results and aiding in query refinement.
This historical context and example serves as a cautionary tale through which to view chatbots. Like voice assistants, chatbots risk simplifying the search and discovery process to a degree that might strip away the richness of traditional search and product discovery.
The key takeaway from the voice search hype is the importance of balancing the pursuit of innovation with the preservation of essential functionalities that make the shopper experience flexible and efficient. As we venture into the realm of conversational commerce, it’s crucial to capitalize on the lessons learned from the voice search trajectory, ensuring that the allure of chatbots doesn’t overshadow the comprehensive capabilities that users expect from search interfaces.
5. Vulnerability and Security Concerns
The introduction of chatbots also brings with it a host of risks and security concerns that cannot be overlooked. Among the most pressing is the vulnerability of these systems to data poisoning and malicious manipulation, which can severely compromise the integrity of the interactions and, by extension, the user’s trust in the brand.
A notable example that highlights this vulnerability was the recent manipulation of a Chevrolet dealership’s chatbot, which was led to generate inappropriate responses. This incident not only caused immediate harm to the dealership’s reputation but also served as a stark reminder of the potential long-term consequences on brand perception and customer loyalty. Such vulnerabilities represent a tangible risk, with researchers indicating that for a modest sum, malicious actors could significantly corrupt the data pool that chatbots rely on, leading to misleading, incorrect, or harmful interactions with users.
The implications of these security flaws extend beyond the immediate interactions. In the broader context of ecommerce, where brand reputation and user trust are paramount, the introduction of a technology that can be so easily compromised poses a significant risk. It suggests that the rush to implement conversational chatbots could inadvertently expose brands to attacks that not only disrupt the user experience but also erode the hard-earned trust that companies have built with their customers.
Relevant reading: A Simple Guide to ChatCPT and Generative AI in Ecommerce
A Better Approach to Conversational Commerce
Rather than doubling down on chatbot interfaces, there’s a better, more effective way to embrace conversational commerce while retaining the strengths of traditional search interfaces. This approach involves integrating conversational AI and generative AI within the search framework to address complex queries and enhance customer education.
Innovating with Conversational AI
Research by the Boston Consulting Group highlights a key expectation among consumers: the most valued feature in GenAI-powered conversational commerce is its ability to address complex, product-related inquiries. This finding aligns closely with feedback from our customers, demonstrating the significance of buyer education and knowledge discovery throughout their purchasing journey.
Coveo is at the forefront of this innovative shift, redefining the role of AI and GenAI in ecommerce search and discovery. By fostering interactions reminiscent of those with knowledgeable sales representatives, Coveo is setting new standards in providing accurate, relevant responses to product related questions. This strategy not only facilitates product discovery but also crucially contributes to educating and empowering consumers, enriching their overall shopping experience.
Take, for instance, a customer asking, “What’s the right paddle board for a beginner?” Coveo’s GenAI-powered search interprets the question and offers a detailed response, highlighting essential attributes like board width and material. Or consider the question “What is the best lens for wildlife photography?”
This level of insight showcases retailer expertise right from the initial stages of product research, engaging customers with precise, informed answers to questions.
Conversational Commerce Considerations
Not all GenAI and conversational commerce applications are equal. The phenomenon of “hallucinations” – where AI generates incorrect or misleading information – poses a real challenge. Caution is advised when considering vendors that leverage LLMs for content generation without a clear explanation of how they ensure factual accuracy.
Coveo addresses this issue with Retrieval Augmented Generation (RAG), a method that mitigates the risk of hallucinations by integrating relevant data during the query process. By grounding an AI model with pertinent information, RAG facilitates the generation of coherent, fact-based responses, complete with citations. Although technically complex, implementing RAG is crucial for the confident application of generative answering on an enterprise scale.
Leveraging generative answering, LLMs, and RAG, Coveo enhances customer education and product research, providing value in a variety of sectors, from home improvement, grocery, and life sciences, to B2B manufacturing and distribution. For example, Coveo can adeptly handle queries ranging from B2C concerns like “I need to build a new deck; what are the steps and what tools will I need?” to B2B considerations such as “What are the disposal considerations for this lubricant?” and “What types of conformal coatings can we source for our new circuit boards?”’
What’s Next for Conversational Commerce?
Coveo is expanding its capabilities with “Catalog Aware Generative Answering,” which links our grounded answering system to product catalogs. This innovation suggests categories and attributes during the shopper education phase, enhancing the potential for conversions. Future enhancements will focus on contextual follow-ups and query reformulations, further advancing the conversational commerce experience.
Dig Deeper
Curious to learn more about what value conversational commerce can bring to your business?