With the consumer shopping journey at the top of every retailer’s agenda, personalization continues to be the hottest buzzword for retailers. Retail tech provider True Fit knows this better than anyone, offering partners the tools to make data-driven decisions for boosting sales and increasing shopper loyalty.
And with new predictive analytics and data science, the company is taking personalization to the next level by leveraging consumer recommendations for an increasingly personalized experience.
“More and more fashion brands are adopting a consumer-first mentality and those that are thriving are using data to put their consumer at the center of their recommendations strategy,” said Romney Evans, cofounder and chief product and marketing officer. “Data alone doesn’t make experiences better, but data is the mechanism for scaling thoughtful experiences that can dramatically improve relevance for consumers. Brands who use data to create experiences that turn browsers into loyal repeat purchasers are doing it right.”
With more data the company will continue to innovate, making technology for retail partners ever smarter. True Fit’s evaluation methodology for recommender systems compares the performance of three different algorithms: the most popular items strategy, collaborative filtering strategy and content-based strategy. Retailers need to consider the performance of all of these algorithms to understand how to meet a fashion shopper’s needs. “The Fashion Genome [a platform by True Fit] put the largest connected data set for fashion at retailers’ fingertips to decode personal style, fit and size for every shopper,” Evans said. “By using a connected data set to power a unique model for every shopper, recommendations become truly 1:1 and not 1:many.”
Finding the perfect fashion recommendation though is a daunting task. Still, through extensive research, True Fit says they have narrowed in on key factors of impact. “Meaningful personalization is challenging, so it takes a while to adopt and refine,” said Evans, “It’s taken a lot of years just to offer really impactful personalization for size and fit on a PDP page. Now that those pipes are laid for so many retailers, I believe they will be able to now apply more style personalization and curation to every other part of the customer journey to more efficiently steer users towards the items they will love and keep.”
True Fit is thinking about the individual. Understanding who each customer is, what will fit and what will be kept goes beyond measurements. “There are many factors to contribute to relevance, including brand affinity, price, region, weather…” said Evans, “but the most important factors relate to a user’s individual style preferences or the combinations of garment attributes that combine to give each garment its unique identity.”
The company learns about consumers when they create a profile on a retailer’s page. There, consumers share favorite brands and fit preferences. Through the retailer, True Fit then additionally analyzes what a consumer chooses to then return or keep. All of this preference data is then used to customize recommendations.
It is a concept, Evans says, that is not unlike recommendations from Spotify or Netflix, which also utilizes a customer’s preference history.
“We’ve found by personalizing the browse and discovery experience with personal style rankings, fit ratings and size recommendations, consumers end up viewing three-times more styles, average order value increases by 19 percent, and shoppers become more satisfied and more loyal, returning to visit 16 percent more often,” Evans said. Findings from True Fit’s research paper “Assessing Fashion Recommendations,” report the personalization aspect to be a central focus to the “fashion user” who expects both unique and relevant suggestions.
Meanwhile, as consumers become more accustom to being given endless selections by retailers, an urgency to provide relevance is also crucial to avoid shoppers’ fatigue. “Making first impressions relevant in terms of style, fit, and size is a make it or break it point for the consumer,” said Evans. To provide recommendations that meet shoppers’ needs, therefore, a retailer must develop an evaluation of consumer’s judgments of similarity to help a shopper achieve goals more efficiently. Showing the consumer, a product with similarity to a “target” item, therefore, can be advantageous. According to Evans, this is especially important in fashion, where consumers more than likely will not search past the first page.
When a retailer is only recommending a brand’s popular items, recommendations that score highly in relevancy, will not include options that are unique to the individual shopper’s preferences and will fall short of a personal experience for the consumer. “AI and machine-learning engines need to be taught to think like humans do,” Evans said. “Algorithms must be built to consider the ways a consumer thinks. Through research, the [True Fit] team is better able to understand which product attributes matter most to consumers when they shop.”
Further, while recommender systems use information about the similarity between items’ features to prompt recommendations, True Fit finds these are inconsistent with how consumers make similarity judgments. In research, True Fit found that some attributes are more valuable than others.
For example, consumers shared that a dress length has more importance above its color and pattern. Therefore, when a retailer looks to make a recommendation to a shopper, retailers need to know something about both the individual item and the individual person who plans to wear it to understand fit. “I think in some cases you have retailers that are using generic recommendation models that use the same methodology to sell washers and dryers as they do to sell clothing,” said Evans, “These generic recommendation models aren’t considering the important details related to personal style preferences, fit, and sizing that every consumer has to overcome to have a successful purchase.”
According to True Fit, a recommendation strategy that performs positively only with prior consumers is a poor fit for fashion data which is characterized by a large majority of new users. Many fashion recommendation systems do not make this distinction into account when evaluating. Also exceedingly important to understand the consumer experience online is knowing when consumers shop online, motivations vary, meaning some shoppers will come to a site with a specific goal in mind and others will be exploring.
“People want to be understood, and served an experience that caters to his or her needs and makes shopping easy,” said Evans, “The recommendations can be used top-of-funnel for search, email, or retargeting, or even in-store to help empower sales associates or enhance self-serve apps.” When a balance is found data can create efficiency for both the consumer and the retailer.
For More WWD Business News: