Digital retailing requires active management of product information, images, and attributes. Managing these consistently and effectively is required to ensure the customer has a pleasant shopping experience, yet most pure play online or multichannel retailers are doing a bad job. The effect is most troubling when filtering the vast lists of hundreds or thousands of products on a website. The filters that are offered for the customer are not reliable. This is particularly true in complex fashion apparel and home products, an area where they are needed most. A recent attempt to select a ruffled bedding set on a name brand ecommerce retailer didn’t end well. They did indeed have a filter for “ruffle”. The results displayed pages of bedding items, about 1/2 of which had a ruffle of some kind. Others were in no way ruffled. They may have an accessory called a “dust ruffle” but that accessory name does not mean the set contains ruffles. The same is true for filters like color. Select purple and the result is mostly purple, however there are mint green, grey, and brown sets shown. Some are questionable, for example a pattern that has a small line. Partially this is the danger of endless isles and marketplaces but that is not entirely to blame. This type of response to filtering is the norm across most websites.
But why does this problem exist? Some retailers believe that they are better off showing the same color ensemble in multiple related colors to maximize exposure. Others leave it up to the eye of the marketing or merchandising associate. Frequently the vendor provides item descriptions that may not adhere to data standards. Truthfully this is hard to manage but required to be successful. If the customer chooses to use a filter to manage the endless selections, they expect that it will be applied correctly. Robust master data management (MDM solutions) and product information management (PIM solutions) are vital to improving this process. Starting with customer experience is critical, but it also has longer term ramifications in merchandising. As more advanced analytics are applied to merchandising processes the basic components of data management are critical. An incorrectly attributed item can have a radical impact on planning. Are ruffles a top seller or is there a different construction that is mislabeled driving the ruffled sales? Is orange trending up or the mint green ensemble? Until the basic data management challenges are mastered forget using much of this big data further up stream.