Fashion Search: Finding the White Whale
I am frustrated. I am frustrated that in the year 2024, in a world where I take driverless cars to run errands, that it requires an inordinate amount of time and effort to find and buy a pair of buttercup yellow d’orsay heels to my liking.
As an avid shopper and fashion enthusiast, I am baffled by where we are with regard to tools that yield desired search results. On the spectrum of fashion search — from wanting something very specific to seeking inspiration — current options don’t meet user needs. Highest priority requirements for me are the following:
- Full breadth of results, not limited to partners and advertisers, within search parameters
- Exclusion filters because often it’s easier to define what you don’t want vs what you do
- High-quality, curated inspiration including complete outfits to fuel discovery and creativity
Nearly a quarter century has passed since J-lo’s Versace dress inspired Google to create image search in 2001. About a decade later, Like.com was acquired by Google. At the time, Like.com was a fashion enthusiast’s dream. Still early in the days of e-commerce, Like.com allowed you to enter a visual (apps today can do this using Google’s Vision API) or text prompt. Your reward: a gallery of visual results where the search results improved as you scrolled and an easy way to compare products. Like.com shut down after the acquisition.
Google Shopping has undergone numerous iterations, including one where search results were primarily PLAs. Unfortunately, these iterations did not prioritize apparel/accessories search. Rationally, the TAM for consumer staples search is much larger than for non-discretionary and the growing threat there is Amazon, but the decision left a sizeable gap in the market. While for many shopping use cases Amazon has become the de facto search tool, it is inadequate for many fashion search use cases. For brands beyond mass market, the appeal is anchored in inaccessibility and therefore incompatible with a presence on Amazon.
When Pinterest launched in the 2010s, I was an early adopter. While it wasn’t a useful tool for search, it was convenient to have a single platform to collect style ideas and inspiration. Initially, my feed was great. It felt inspirational, new ideas from a curated set of fashionistas I followed. Unfortunately, the feed became more and more cluttered with pins from users I did not follow; users Pinterest erroneously believed I would want to follow. Despite thousands of pins on my style board, the algorithm could not define my style. As a tech executive, I could see that the myriad of factors, often quite nuanced, that define fashion style and preference were not incorporated into the feed algorithm. During a session, pinning an image meant I’d see more images in my feed that were extremely similar to what I’d just pinned. Therefore, my feed became repetitive - the opposite of inspiring. Of the times I’ve attempted to search on Pinterest, the results have been disappointing. I’m looking forward to seeing if LLMs can be created and trained to solve these complex issues.
During the pandemic, I became a user of The Yes. This was the first time since Like.com that I felt a platform targeted users like me. Fashion enthusiasts! Style connoisseurs! For those willing to spend time — no, loving to spend time — shopping! It delivered adequately against both ends of a fashion search user’s intent spectrum — most useful for finding a specific product while offering a dose of inspiration. When I searched for a specific product type, I was presented with results which improved over time because the platform allowed me to convey my preferences through yes/no buttons. The machine learning actually worked — my results improved over time and became more and more useful although they appeared limited to brand partners. To the same end, I could follow specific designers and brands which helped me not only hone my search results to preferred designers and brands, but the algorithm suggested others based on my preferences. Some of these were brands I knew of, but had forgotten — huge win for both user and brand. Some of these were brands new to me, even as a close follower of the industry — also a huge win for both users and brands.
Moreover, as someone who misses the curation of fashion magazines (yes, it’s freeing to not have fashion overlords defining trends, but at the same time guidance based on my distinct style and the zeitgeist is very welcome; RIP Polyvore; Instagram and TikTok aren’t really it for this use case), The Yes had one of my favorite fashion editors, Taylor Tomasi Hill, as the creative leader who provided guidance. While I don’t dress like her, she is inspirational because she’s highly creative with good taste - offering new, of-the-moment, ideas for outfits and unique flourishes from which I could pick and choose. While The Yes’s email formats were not ideal for me (lots of flashing imagery resembling a strobe effect), the content was good and more importantly, necessary to supplement the user experience while offering brand partners increased visibility. The Yes was acquired by Pinterest in 2022 and shut down.
We need a search engine for fashion and style enthusiasts and over time, for the general populace. Designed to meet the needs of those of us who have a distinct, defined style; who are hobbyists or professionals in the fashion design and style realm. Supported with content marketing that seeks to expand horizons, feed imaginations, aid discovery. Leveraging the recent advances in machine learning to truly comprehend nuances that define originality, revel in the complexities of individualism, synthesize an infinite universe of styles, produce a unique experience that inspires and expands self-expression and finally, deliver accurate search results, not limited to brand partners, that users seek to help bring their fashion creativity to life.
