Merchandizing – Enabling Product Discovery in a Digital World
Picture this. You are at your favorite retailer’s e-commerce site and your keystrokes announce why you’re there – you’re looking for “a sort of traditional prom dress in a darker shade” or “frayed light blue jeans with a lot of pockets and chrome buttons.” The output? A set of irrelevant (or worse, zero) search results leaving you scratching your head and abandoning the site.
In a digitally connected world, where channel, device, app and site hopping is the norm, the shopper is searching, experiencing and buying products in drastically different ways. Many retailers are struggling to align product information to how the shoppers experience the product and consume it. In fact, many still stick to outdated ways of organizing information around their product selection and bulk buying priorities.
Digital design puts the human experience and empathy towards each individual customer at its core. This means understanding his or her aspirations, journeys and preferences, and then being a companion in their quest for product discovery. In a world with near-infinite choices, the capability of narrowing down options based on affinity and preferences plays a significant role in sales conversions.
To better understand how to enable product discovery in the digital world, let’s first take a look at the evolution of retail product data - from master data management (MDM) to product information management (PIM) to attribute-based merchandizing. From there, newer technologies like analytics and machine learning can help retailers crack the code of lost sales in the digital world as a result of poorly defined attribution.
From Product Information to Product Discovery
Right from the early brick-and-mortar days of retailing, buyers and merchandisers have acquired and owned product information for their categories. Typically, this data was born out of discussions between buyers and suppliers. The resulting catalogs used to focus on a few core product attributes like size, color, and fit, in the case of apparel, and were considered sufficient to aid the buying decision. However, oftentimes information trickled down poorly to stores where the focus was largely on pricing or promotions and not on product information. For instance, flavor and package size often determined sourcing of packaged food items instead of nutritional content.
This limitation was reflected in the product MDM systems used by retailers early on, which were typically a data store within the merchandizing systems. These systems circled around governance and provided limited options to expand attributes of items based on categories. Even while these systems claimed a capacity of storing hundreds of attributes, it was often poorly collected and disseminated. As a result, the ability to derive cross category or product relationships were non-existent. This also limited the amount of customer-product insights one could derive since the buying motivations could not be tracked accurately. Limited assortments and even more limited product information meant that shoppers made purchase decisions based on touch-feel-try or peer/expert recommendations.
The e-commerce explosion changed the way retailers and customers perceived product information. Shoppers needed help narrowing down choices, giving rise to more nuanced product information as a potential differentiator and creating a world of faceted search. To achieve better conversion rates, assortments had to be complemented with the ability to create multiple catalogs and ensure speed-to-market in terms of product information, campaigns and promotions.
PIM systems evolved by adding capabilities to support multiple user groups as well as expanded attribute sets to best-suit customer searches. Suppliers were empowered to upload product information, which could be cross-referenced across categories for clustering and creating promotional events. Marketing could turn these into dozens of tailored catalogs and target promotions based on geography, ethnicity, seasonality and various other factors that influence purchase decisions.
However, PIM was still top-down, inside-out, push-based and relied on defined pathways which the shopper had to navigate through. These fundamental limitations were exposed time and again when shoppers started seeking products outside of the pre-set journeys, seamlessly moving in and out of the retailer’s confines.
The future belongs to those who use product information to aid discovery through any means possible, but who also keep the shopper at the core. The fast-evolving capabilities of digital and machine learning technologies mean that retailers can gain a wealth of information about the customers through their search history and convert it into actionable insights. For example, when Maya complains on Twitter about not having silver-plated buttons on a gown she bought, or when Peter looked for a pair of jeans with slanted pockets and white stitches, it’s an opportunity to ensure button-type, pocket style and stitch color become exposed, searchable attributes.
One of the leading department stores in the U.S. observed that ‘heel arch support’ was one of the most searched features in the footwear category even though it was not a defined attribute in their e-commerce site. While several product variants existed, those never showed up during the search. Similarly, our grocery clients have seen the impact the ‘fitness revolution’ has had on the way consumers shop for their food and beverages. Concepts like ‘gluten-free’, ‘organic’, ‘vegan’, ‘trans fat-free’ and ‘sugar-free’ have become mainstream enablers in choosing products. Systems that evolve to not only suggest products based on affinity but alert purchases that do not suit the customer profile, such as allergy-based alerts, will experience a surge in adoption.
Staying Relevant in an Attribute-Driven World
Product information in the digital era continues to be incredibly complex. First, there is the proliferation of categories and stock keeping units (SKUs), providing a much greater breadth and depth of assortment. Moving forward, diverse sets of relationships need to be built across traditional definitions of catalogs and promotions (e.g. bundles and ensembles). Retailers need to add value to the shopping process in order to drive conversion.
Second, beyond text, we also now have images, videos, reviews, Q&A forums, celebrity endorsements, educational content and various other forms of user-generated information flooding the internet, created by passionate and vocal niche category-level influencers. Retailers face the challenge of extracting, validating, prioritizing and leveraging information from these incoherent data sources for the purposes of contextualizing content to each shopper. Leveraging these sources to achieve taxonomical relevance, define closest substitutes and plan store layouts will make a huge difference.
Third, click-stream and user-generated content must be tapped for insights on consumer behavior and to enrich product data. For example, Amazon.com, which has always been a ‘mentor’ for retailers to learn from, created an authenticated review and Q&A forum that buyers rely on.
Finally, with marketplace models gaining adoption, the consistency and completeness of data provided by suppliers is coming under strain. As a result, there is an opportunity to develop tools that can parse through lengthy unstructured product descriptions, derive suitable attributes and enrich poorly attributed product data.
As product data and its attribution evolves, the one thing we can all be sure of is that the era of setting product information on a one-time basis during the life of a SKU is truly over. Anyone not infusing a continuous learning culture into PIM with machine learning and predictive analytics runs the risk of being completely bypassed in the era of pull-based digital product discovery and lifestyle driven shopping.