As more apparel shopping moves online courtesy of COVID-19, accurate and thorough product tagging could mean the difference between a consumer finding what they want or leaving empty handed.
E-commerce is known for its convenience at the point of purchase and delivery, but product discovery can be a challenge as shoppers need to scroll through numerous product pages. As a solution, e-tailers have added filters and search functions, but keywords can be limiting when it comes to fashion searches since verbal descriptors of style traits can vary.
Garmentory, a marketplace that acts as the e-commerce arm for more than 1,000 independent boutiques, experienced this difficulty firsthand as it tried to manually tag its 100,000-plus styles, since one merchandiser’s “puff sleeve” could be another’s “balloon sleeve.” Overcoming this lack of standardization, Garmentory is the latest company to partner with technology firm Donde Search to automatically tag merchandise using computer vision and artificial intelligence.
“Marketplaces like ours present a ‘paradox of choice’ to the shopper—while the vast assortment offers plenty of choice, finding that hidden gem matching exactly what they are looking for can present some challenges,” said Sunil Gowda, co-founder and CEO at Garmentory. “With Donde’s product tagging combined with their visual attributes-based product recommendation, we are able to offer a delightful shopping experience to find that perfect item without requiring [consumers] to sift through thousands of items.”
Compared to visual search solutions that look for broad similarities between images, Donde’s technology takes a more granular approach by identifying thousands of possible attributes in a product.
Transferring product tagging from a manual process to an artificial intelligence-driven one creates a singular visual language for categorizing merchandise. It also expands on the number of product attributes cataloged, creating more data. On similar projects, Donde has added about 20 attributes per product category. Whereas manual tagging rates for many features don’t reach 50 percent, Donde says its solution has a coverage of 95 percent per characteristic.
Gowda says using AI has lowered the cost of the review process and has shortened the time it takes to make merchandise available for sale on its site.
Aside from facilitating searches, Garmentory is also using Donde’s technology to power product suggestions by finding similar items. Personalization features are often powered based on clicks, requiring an item to be viewed by others before it is served up to a shopper. Comparatively, Donde’s recommendations are based on matching product attributes, allowing items to be linked even when they are new.
“As Garmentory grows by onboarding new boutiques and offering a larger assortment of items, it becomes imperative to fully automate the product classification process,” said Gowda. “Similarly, providing a best-in-class shopping experience on the product discovery front will be critical to our continued growth and success.”
Going forward, Garmentory will also be using Donde’s Style DNA solution to discover what attributes are resonating with its customers. Rather than just showing which products were most popular, this provides a more top-down view of trending styles. Eventually, this can be shared with merchants at Garmentory’s partner boutiques to help inform their buys.
“Due to the unique information that we provide in every item and with every action that the shopper is doing, Garmentory can actually get more insights about the most popular styles in terms of attributes that people are searching for, viewing and eventually purchasing,” said Eli Cohen, chief product officer at Donde Search.
Donde’s other clients have seen their average order values and conversions rise by up to 20 percent by implementing the tech firm’s AI Visual Discovery technology.
With e-commerce picking up and more merchants and consumers moving online, Donde sees a greater need for its technology.
“The problem that we’re solving that we call ‘shopstration’ is becoming even bigger because there’s so much variety and more people are going online,” said Liat Zakay, CEO and founder at Donde Search. “So I think that now the value of Donde AI Visual Discovery is even bigger, and more importantly helps the consumer improve the shopper experience and helps retailers increase conversion, which is very critical and significant in these challenging times.”