Just 3 weeks ago we discussed AI as a megatrend in e-commerce right here. This week, we’re taking a deep dive into the topic of shopping assistance. For over 15 years, Comsysto Reply has been implementing complex e-commerce solutions for clients across all industries. At the same time, we track and analyze relevant trends in our e-commerce lab to ensure we are ready for tomorrow's challenges today. We share our insights here—every Tuesday!
E-commerce and LLMs
For a good two years now, almost every industry has been discussing how the use of LLMs (Large Language Models) is opening up entirely new fields of application or revolutionizing existing ones. Especially where there is intensive interaction with end users, we are already seeing massive changes. The most prominent example of this is ChatGPT, the fastest-growing app of all time (Forbes). And to stay within the e-commerce sector: at Amazon , the “Rufus” shopping assistant has been available to some customers in their mobile app for a year now (Amazon). It is no wonder, then, that most experts see a virtual advisor for online shops as one of the most important trends for 2024 (e.g., Geert Leeman from SAP in his outlook from just over a year ago) as well as for this year (e.g., E-Commerce Germany).
But why is this so relevant?
We have last week already discussed this in more detail: there is no way around marketplaces! And with the integration of products from partners and third-party sellers, product catalogs are growing significantly - making it increasingly difficult for users to navigate them effectively.
Searching for and selecting an item can easily lead to frustration and, ultimately, to a "drop-off" (Shopify), which we all want to avoid. In a world where users are increasingly interacting in a "conversational" manner—using natural language—we can assume that this trend will not stop at online shops. In practice, this means that users will talk or chat with a virtual assistant, while the appropriate catalogs and filters for the search query are selected in the background. Paired with data from previous orders and a comprehensive knowledge base for the respective product category, it is possible to guide users directly to the products they are looking for without unnecessary interactions.
But that's not all! As soon as the customer has found the item they are looking for, the next hurdle arises: How should they choose between many items that look similar or are even identical at first glance? AI can also help with this problem, often referred to as "analysis paralysis" (Shopify): by asking the right questions, which are derived from the search history and the conversation held so far. Similar questioning techniques can be used as those employed by human sales staff (e.g. business-wissen.de), who, for example, naturally guide users to product reviews.
Similar to a huge product portfolio, the user is often overwhelmed here too! Yes, reviews are an asset for any online shop (Trusted Shops), but they can also quickly overwhelm customers, as shown not only by academic studies (e.g. UNLV). That is why it is particularly important to provide the user with only essential information whenever possible. This could be a summary of all reviews or a selection of the most important reviews for this specific query. A problem perfectly suited for LLMs…
How can I integrate a shopping assistant?
There are various ways to integrate artificial intelligence for product advice. The easiest starting point is to expand the search bar, which is available practically everywhere, by connecting an LLM so that it can also understand natural language. You can find out much more about this from the AI personalization experts at Bloomreach read more.
Using a chat takes things a step further – here, too, the existing search and filter functions are used, but the user now engages in a truly interactive conversation to refine their selection and the products displayed step by step. This is not only practical, but also the perfect opportunity to ask for more information or engage in cross-selling – just like in brick-and-mortar retail.
For those who would like to get even more technical: in our e-commerce lab, we are currently working on an article for our engineering blog on this very topic. How should an AI agent be structured, and how do I integrate it into my online shop and system landscape? More on that here soon!
And if you would like to simply try out the experience from a user's perspective, you certainly won't be disappointed by the AWS demo page on the topic of shopping assistants for the apparel industry!
That’s it for today – and here’s what’s next
For once, we aren't closing today with a look at the next issue of this newsletter, but rather at the calendar. The E-Commerce Expo is taking place in Berlin on February 19th and 20th – will we see you there?
See you next week – hopefully in person!
Christian & the Comsysto Reply e-commerce lab