Fast Fashion Online: It’s All About Product Data Quality
Zara changed the retail industry forever by introducing fast fashion. The business model depends on getting up-to-the-moment fashions into retail sites at speeds that were previously unattainable. Zara competitors must offer new designs, based on real-time customer preferences, and drop-ship them into inventory — just in time — all while urging customers to act quickly if they want to find optimal choices. Any delays across this sophisticated and data-driven supply chain can jeopardize sales and compromise a brand’s ability to compete.
As an increasing proportion of fashion sales move online, the fast-fashion model presents an additional layer of challenges. Success online not only depends on the ability to refresh inventories quickly but also on the quality of the underlying product data. Successfully addressing the product data challenge is the key to success in fashion e-commerce.
The product data challenge
A typical clothing item has dozens — sometimes hundreds — of individual product attributes, ranging from basic information such as color, size, and fabric type to more complex characteristics such as the item’s country of origin, how it was manufactured, and whether it meets fair trade standards. Every time a fashion brand updates its inventory, hundreds of product data attributes across thousands, even millions, of products need to be managed and updated as quickly as possible across the distribution network, both in retail outlets and across e-commerce channels.
For fast-fashion clothing items with short shelf lives — sometimes as little as 15 days — this problem is enormously amplified. Any lost time involved in getting products tagged and available for sales channels translates directly into lost revenue. Manually entering product data, or relying on outdated tools such as Excel, can end up taking longer than an item’s entire shelf life. For fast fashion to work in this rapidly moving context, new solutions are critical.
The status quo
Notwithstanding the hyper-speed of today’s fashion and e-commerce, the majority of fashion product data is still managed and updated using spreadsheet programs such as Excel, the favorite hunt-and-peck software program of recently minted 1990s MBAs. But handling data in this manner is becoming increasingly obsolete given the fashion industry’s evolving digital transformation and the expanding number of channel options available to today’s brands. Manual processes are not just slow and expensive; they are also prone to error and cannot scale with increased volume.
Failing to move beyond the limitations of manual product data processing can negatively affect businesses in additional ways. According to eConsultancy, 56 percent of online consumers abandon purchases because of lack of information about the product or delivery. If products are not connected to accurate, up-to-the-minute datasets, there can be major discrepancies between what shoppers search for and the products that surface, potentially resulting in lost sales, negative customer experiences, and reduced brand loyalty. Retailers are increasingly realizing they need to find innovative solutions to condense the amount of time and labor it takes to get products — and their associated data — to market.
The power of artificial intelligence
One way for brands and retailers to establish and maintain competitive advantage is by embracing new and scalable solutions that harness the power of artificial intelligence. These new practices are driven by advances in machine learning, applied to proven industry frameworks, allowing for automated creation and management of product attributes.
AI-enabled product data management is a critical element in this new world of move-fast-or-die-trying fashion manufacturing and retailing. Computers can categorize and process information about size, color and fabric type, not to mention countless other product attributes, exponentially more quickly than humans, with comparable or greater accuracy. But fully AI-enabled systems can also do things that are nearly impossible for humans to do, such as near-real-time updating of millions of SKUs, rapidly examining search patterns, or analyzing vast numbers of shopper queries. An automated system can scan purchasing data and determine almost immediately that a bead-laden turquoise-colored blouse is dipping in popularity, while a puffy-sleeved beach shirt is flying off shelves.
For fast-fashion retailers that are determined to compete in the fast-paced online world, it is critical to digitize supply chains and be capable of analyzing real-time sales data to understand where customers’ preferences are trending. This can best be accomplished using the latest AI technologies: A national retail chain with thousands of stores, for example, is now able to process data about several million products in under an hour using a properly configured AI-powered system. This is an achievement that was not even remotely possible only 12 to 18 months ago.
The most powerful benefits of using machine learning in product data management come from the interaction between product information and its use in the consumer search and discovery process. Advanced machine analysis of customer queries and purchase patterns can help companies create better and more detailed product descriptions, which in turn allows items that otherwise may have been overlooked to be discovered, evaluated, and moved through the sales process.
Fast fashion is not for the faint of heart. Its success is based on a huge number of factors, some straightforward and others that can take years to master. But if the world of e-commerce has taught us anything, it’s that today’s incumbents can become tomorrow’s roadkill. Just look at America’s suburban shopping centers, all too many of which are now littered with empty parking lots and shuttered doors. Brands and retailers of all sizes can rebalance the deck to their advantage by embracing these new AI-powered product data tools and processes sooner, rather than later.
Divyabh Mishra is founder and CEO of CrowdANALYTIX.