Curated Clothing: Giving Consumers What They Crave Online

According to the Personalization Consumer Survey, more than half of consumers surveyed say they spend more money with retailers that make product recommendations based on browsing or buying behaviors. Forrester's "Future of Shopping" report by Sucharita Mulpuru recently identified personalization as one of six opportunities for retailers to address as they aim to grow business. 

In fact, Forrester predicts that the most near-term opportunity for personalization technologies lies in the apparel industry because, with such inconsistency across brands, fit remains a continual problem.  Mulpuru writes, "[This] could represent a significant improvement in the apparel and footwear industry by reducing waste and better matching shoppers to their needs. They also can ultimately improve the sell-through of apparel."

It's about having insight into what a consumer may want even before even they do, and allowing them to purchase it quickly and easily — with the confidence that they won't be back at the post office making a return a week later. Shoppers no longer want to browse through hundreds or even thousands of items, compare pros and cons of each one and read reviews before placing an order. They want a curated experience, one where the best-fitting items find them.  To get this right, apparel retailers have to focus on two key elements: data and algorithms. But here's the trick — it must be based on the right data and algorithms, and when that happens, personalized e-commerce will make it possible for retailers to sell to a much bigger audience based on personal preferences.

Not just big data, but good data. Perhaps it's something we're all tired of hearing about, but data still is the single most important element of creating a curated e-commerce experience. But data is messy, and it's not all created equal. Specific to fashion retail, garments are all related to each other in some way. Just because the data says a garment is a dress doesn't mean every shopper looking for a dress will want to see it. What is the silhouette? Color? What material is it made of? What price range does it fall into? It's important that garments are understood in relation to each other — organized into a sort of hierarchy based on meaningful and unique attributes.

The data needs to be able to very specifically sort and classify garments based on all of their attributes in order to understand how to prioritize the sort for each customer. Some of the most important data to creating a useful and relevant recommendation is user preference data. Just because a shopper has a 17 inch neck doesn't mean he likes to wear shirts that are 17 inches in the neck. He may prefer his shirts loose and favor an 18 inch neck. Shopper preference trumps all, and must be considered in relation to the attributes of each garment.

What pairs best with good data? Algorithms. This may not be the sexiest topic, but without the right algorithms, all of that data is useless. They're the backbone of making use of the billions of data points available — even if retailers don't have access to all the data they want. They must have a graceful way of falling back to having less, imperfect or inaccurate data. If an algorithm can make a recommendation only based on the perfect set of inputs, the result may be no personalization at all. Despite all attempts to get good and consistent data, there are going to be things that don't make sense. The right algorithm will still function, even if most of the data doesn't fit. And for apparel e-commerce, algorithms must consider many more factors than CDs, for example, which don't have the extra dimension of size and color, beyond just the product name.

Like a good sales associate, online apparel retailers can ensure their shoppers only see items that will fit and flatter – creating a quick, convenient and satisfying online shopping experience that inspires confidence. In addition, retailers can increase sales of a specific garment, perhaps something that is overstocked, by making sure the most likely purchaser sees it. By identifying which items each shopper has previously purchased and keeping this information in combination with rich data about the user and the garments, retailers can serve personalized recommendations for garments a shopper is likely to buy (and love) the next time.

While there are many new services trying to reinvigorate fashion sales, the only way to really do it at scale is with the right data and algorithms. There may be a specific consumer that will always want to have her garments pulled by an individual stylist, but to reach the broadest audience with the most personalized experience, it all comes down to what data you have in hand and how it all comes together. And the thing with data and machine learning is… it gets better with scale and with time. So the more people use it, the better the recommendations become.

Romney Evans is co-founder of True Fit, a data company that provides sizing recommendation information for fashion retail companies. 

This ad will auto-close in 10 seconds