Introduction
Recommending products via a chatbot is a valuable feature for any business, but it’s not the end goal. Companies aim to not only recommend products but also to boost customer engagement and drive conversions through these interactions. A highly beneficial addition is enabling users to query for products with discounts as it can significantly enhance the user experience. In ExpertGuru, we have implemented this feature, allowing users to search for items with discounts and filter products above specific discount percentages. This innovation provides a more tailored shopping experience, attracting budget-conscious consumers and fostering customer loyalty.
High-Level Flow Diagram
Challenges and Solutions
Variants with Different Discounts
One of the primary challenges we faced was handling products with multiple variants. Each variant, such as different colors or sizes, can have distinct discount rates. This complexity made it necessary to manage and display discounts accurately for each variant.
Solution: To address this, we extracted detailed information about all product variants from Shopify. We then integrated this comprehensive data into the metadata of our Elasticsearch database. Utilizing a sophisticated language model (LLM), we were able to accurately extract and process discount information for these product variants. This approach ensured that users receive precise discount details for each variant, enhancing their shopping experience.
Incorrect Discount Filtering
Another significant challenge was the inconsistency in the results when querying for discounts greater than a specific percentage. These inconsistencies arose from the format of the input data provided to the model, leading to inaccurate filtering.
Solution: We tackled this issue by modifying the format of the discount-related input data fed to the model. Additionally, we employed prompt engineering techniques to refine the queries and ensure the model accurately understood and processed the discount thresholds. These modifications helped in providing users with consistent and reliable information when searching for products with specific discount levels.
Inclusion of Non-Discounted Products
Occasionally, non-discounted products were included in the results for queries specifically targeting discounted items. This not only confused users but also detracted from the effectiveness of the discount search feature.
Solution: We implemented dynamic filtering mechanisms to ensure that only discounted products are included in the query results. This filtering process is dynamic, adapting to the specific requirements of each query. By focusing only on relevant items, the system effectively re-ranks the products based on the user's query and their conversation history, providing a more accurate and satisfying shopping experience.
Conclusion
The enhancements made to the ExpertGuru chatbot, particularly the ability to recommend discounted products, have significantly improved its product recommendation capabilities. This feature is especially beneficial for engaging budget-conscious consumers, who are often looking for the best deals.
By offering precise and relevant discount information, the chatbot helps to attract and retain these customers, potentially increasing conversions. For merchants, this means not only a boost in customer engagement but also a more efficient way to target and appeal to economically minded shoppers. The ability to search for products by discounts makes the shopping experience more enjoyable and efficient, ultimately leading to higher customer satisfaction and loyalty.
