Navigating the New Era of Digital Commerce with Generative AI

Exploring the untapped potential of generative AI in eCommerce

The applications of generative AI through the digital commerce landscape are vast yet so far, mundane. That’s because a large concentration of human capital in this industry is allocated toward the mundane: historical reporting, ‘fixing’ of publishing issues, website coding and design, writing copy, competitive analysis, campaign optimization, price benchmarking, and the list goes on. What do all of these things have in common? The requirement to capture, process, analyze, and act on data. Bean counting to a large extent. Not just any data but the data that moves the needle on brand or retailer strategy at precisely the right point in time. It’s harder than it sounds and there’s a ton of fat in the process.

Despite advances in machine learning and AI, humans still perform the majority of this work. Retailers and brands — especially legacy ones with established supply chains, economies of scale, and sustainable margins — are still figuring out how to restructure their organizations: people, process, and technology, to take advantage of advances in technology that are already here. The reorganization of decision making brought on by these capabilities requires connection back to the real world supply chain. It sounds obvious and it is, but this is the challenge, especially for organizations who operate like oil tankers. The more agile the connection, the greater the edge. For better or environmentally worse, this is exactly what Shein has won at, so far — capturing a significant edge while increasing negative environmental externalities that they have largely had the runway to legally do.?

With generative AI we can move to a framework of optimization across factors such as pricing, assortment, search, product copy and creative, that is incredibly dynamic given the enhanced self-learning nature of this technology. As the retail world continues to shift toward programmatic and real-time changes, we need to capture insight with greater immediacy. This is important because as shopping assistant technology becomes more prevalent, and incentives less perverse, the pace at which updates need to be executed to win the sale with the shopper increases.?

AI in Action: Three Real-World Applications of Generative AI in Today's eCommerce Landscape

Product Name and Attribute Identification

In the first case, leveraging both real-time prompt tuning and AI, in our case what we refer to as ‘Sherlock AI’, a proprietary tool, we generate the most suitable product names and attributes in line with real-time market trends. ‘Real-time market trends’ refers to attributes such as products, styles, categories, colors, among a multitude of other factors, that have been identified as having a higher ‘probability of success’ in the marketplace based on a ranking of where these attributes lie in a trend cycle. Drivers considered include search ranking, SEO trends, and inventory levels, to name a few. This process is fueled by data scraping methods, allowing for the capture and analysis of intelligence on top-performing similar products, ensuring that the product names and attributes generated reflect shopper preferences and current market realities.

AI-Powered Chat Assistants

In the second case, AI-powered chat assistants provide merchants with data on trending topics derived from the past year. Built on a fine-tuned Large Language Model (LLM), the chat assistant delivers insights into trending attributes, review summaries, and more, enabling informed decisions. The key here is the way in which the intelligence is surfaced to the merchant. Why does this matter? Typically, merchants and buyers at enterprise-level brands and retailers are stretched over large product portfolios that can include multiple categories. The ability to process, let alone act upon, optimization opportunities across the portfolio has been constrained by the bandwidth of humans? — by the ability to bean-count, understand, and execute upon that knowledge, in time. AI-powered chat assistants dramatically increase the bandwidth of merchants — not necessarily replacing them, rather, increasing their leverage in understanding and execution across the product portfolio.

Special Applications for Third-Party Seller Networks

A third example involves integrating generative AI capabilities into third-party marketplaces relying on third-party seller networks. If managing elements such as product copy, imagery, taxonomy, reviews, and more, is hard for brands and retailers focused on their own house — doing it with third-party sellers introduces a whole new matrix of complexities. These capabilities help drive a cohesive consumer experience across all facets of the retailer's platform while empowering third-party sellers to maximize their market potential within the network.?

The Impact

Real-world results of these applications include predicting demand and trend attributes for a large mid-market U.S. fashion retailer with the insights being used to augment manufacturing orders mid-cycle. This resulted in a 20% reduction of stock that moved through to markdown phases, increasing product sell-thru at full margin, and in turn, profit. Similarly, applications for pricing arbitrage. One luxury brand acted upon an opportunity identified in China between its line of handbags and that of a key competitor. The pricing spread wasn’t captured via traditional analytics and analysis alone. The result was an overnight 60% increase in margin. The range of upside depends largely on the category dynamics and margin economics already in play. For luxury brands, the upside per SKU can be very significant where margins are already wide. For grocers, the upside by SKU will be razor thin, 0.5-3% we typically observe, a function of the tight margins in this vertical; however, when multiplied by the sales velocity, the gains can be lucrative.

These advances offer all retailers and brands the confidence, tools, and insights to effectively navigate a dynamic landscape, make data-driven decisions, and ultimately, drive profitability. The challenge is the reconfiguration of humans in the organization around the tools. This is easier said than done, especially for enterprises tied to a quarterly earnings cycle.

Curious about the potential of generative AI for your brand? Explore how Intelligence Node is transforming digital commerce with generative AI on our digital shelf platforms. Schedule your demo today!