Welcome to our new column, Dynamic Voices, a regular series on XP² featuring timely perspectives on personalization from thought leaders within our network. For our second edition, Jennifer Petry, Fashion Subject Matter Expert, shares how optimizing your product feed for trends can improve overall personalization efforts.

Despite soaring labor/raw material costs and persistent shipping delays, fashion companies have been remarkably resilient over the past few years. But to truly thrive in 2024, they’ll need to meet users where they’re at this season. For one, steep price tags on everything from groceries to transportation may lead to more understated and less ostentatious clothing purchases. Moreover, consumer demand for ethical products is rising, and new sustainability guidelines could drive a swifter transition to industry-wide supply chain accountability. While retailers need to navigate these new macroeconomic trends, much of the regular playbook still stands: For many, the desire to stay fashionable will supersede any other concerns.

Personalization is one of the best tools retailers have for jumping on trends—whether seasonal or industry-shifting—as it enables them to serve decisive customers who know what they want, as well as engage those who may want some assistance with their purchase. By using tools like affinity allocation, retailers can serve a relevant, hyper-targeted banner based on a site visitor’s preferences, which can boost conversions, brand loyalty, and engagement. Take the two banners below, for example, which are personalized and served to different customers based on their user affinity, comprising interests, behavior, and intent. 

Hero banner aimed at shoppers who’ve shown a preference for the boho style, spotlighting long, floral print dresses and wide brim hats.

Hero banner highlighting new arrivals under the Y2K style, targeted at those who are looking for the trendiest clothing this spring.

While a powerful tool, personalization does require an up-to-date foundation of product data to serve users relevant content and timely product recommendations. In the world of TikTok and Instagram, fashion trends cycle through at breakneck speed. That means there’s a real risk of accumulating product attributes for styles that are either irrelevant, inconsistent, or no longer en vogue. You likely have more information about your products than you currently believe, so now is the time to get some spring cleaning done. Identify product attributes that are not specific enough to be of use, then create new, actionable ones that align shifting consumer preferences. Your users—and key stakeholders—will notice the difference.

Tidy Up Your Product Data for Optimized Recommendation Results

After a strategic acquisition left home24 with over 300,000 products, Gianluca Randisi, the brand’s Chief Product Officer, decided it was time to complete the incredibly technical, time-consuming, and complicated project of streamlining the company’s product feed. It was, after all, the only way to improve recommendation quality and ensure the data feed’s integrity as the program scaled. By doing so, home24 greatly reduced costs and tripled its share of revenue. That’s just one example of the transformative impact a clean product feed can have on a retailer’s bottom line.

So how did home24 do it? To get similar results, you should clear your product feed of duplicate, unused, irrelevant, or inconsistent product attributes. For example, a site visitor may be looking for an academic blazer to go with their tennis skirt—both of which should fall under “athletic prep”—but may come up short in the product recommendations widget. Another could end up sifting through leggings or polos tagged “athletic” or “prep,” but completely miss the head-turning items that embody such trends.

Next, you should check whether your keywords are referencing useful differentiators, such as color, material, specific campaigns, fit, sale, etc., rather than repeat terms in different languages, overly specific descriptors (“belt on waist”, for example), fashion brand marketing terms (like the term “mystic blue” rather “blue”), numbers (“length 7.5 in”) or percentages (75% viscose). From there, focus on making updates to your product feed by lining up keywords and attributes under a specific trend, personality, or style.

Foster Cross-Department Collaboration

Rarely is product feed cleaning a one-person task. Product development, buying, and marketing teams should meet internally twice a year to identify trends, colors, pattern styles and product types they might not have in their feed yet. Be sure to get on the same page about whether existing product attributes suffice, or new ones are necessary. Once brands find a structure for the process and build it into their seasonal cycles, it will become easier for them to maintain. With these elements in place, retailers can move on to creating a mapping file.

Product Feed Mapping for Personalization

Let’s create a mapping file for two nascent fashion trends: the rise in Y2K throwback fashion and a growing interest in sustainable clothing.

The Y2K style, centered around retro fashion from the 2000s, is marked by colorful clothing, eye-catching accessories and revealing tops (think midriffs and baby-sized t-shirts). Since it’s a polarizing trend—some may find it too garish or off-putting—it’s the perfect candidate for eCommerce personalization. But you can’t personalize your offering if you don’t have information around the Y2K trend in your product feed.

So how do you go about doing that? In your internal meetings, come up with a list of at least 4 product attributes that fall under the Y2K trend. While a personalization team may not have the expertise to do this on their own, this is a great opportunity for fashion and merchandising experts to lend their voices and guidance.

Below is a basic list to help you get started:

Likewise, you should build a list of product attributes for sustainable clothing that includes not only materials and patterns but popular certifications as well. Below is another jumping-off point for your marketing and product teams:

For sustainability, you would benefit best from putting your keywords/attributes in a “True” or “False” column, through which clothes can be identified as either sustainable (True) or not (False). That way, a product that has at least one attribute from the list will produce the value “True,” and can be dynamically served in a corresponding recommendations widget.

Consider using an additional True or False column for what’s trending as well. Clothes that fall under the Y2K trend, for example, would identify as both Y2K (True) and trendy (True), and be served as recommendations to those who have an affinity for the latest in fashion, as well as those who have a specific interest in Y2K throwback apparel. Note, however, that while Y2K is “trendy,” it might not be equally trendy for all demographics. By adding modifiers to account for different demographics (like Gen Z or other user groups), you can drill down specific audience segments and serve relevant trends to different groups over time. For instance, while Gen Z is into Y2K fashion right now, they’ll probably be interested in another trend down the line. Maintaining high-level data on trends and demographics will help keep the core of your product feed categories timeless and adaptable.

An Optimized Product Feed Means Improved Personalization

While cleaning up product data can be daunting, neglecting your feed can render an entire product database unreliable and in turn, unactionable. The effort is usually well worth the investment. Consumers reward those who get their product feed—and by proxy, personalization—right. Nowadays, 72% of shoppers expect the businesses they frequent to recognize them as an individual and know their tastes. When brands are built on relationships, rather than transactions, they remain loyal and engaged for decades. And ultimately, the positive impact of this approach on conversions, product sales, and customer retention extends across the entire company—not just the personalization department.