Dive Brief:
- Etsy introduced a new browseable search experience Wednesday to its app, launching curated collections that draw on AI and human intelligence.
- Experts first make the collections, which draw on trends, aesthetics and occasions, then Etsy engineers use machine learning to expand around 50 hand-picked listings to about 1,000. Etsy then uses large language models to ensure the collection is “aesthetically cohesive, represents a variety of products, and meets our standards for quality,” Nick Daniel, chief product officer of Etsy, said in the announcement.
- Shopper activity, such as recent purchases and previously viewed items, determine the trends a user sees. What a shopper interacts with most is highest on the feed. Collections are based on trending aesthetics like “modern farmhouse,” “island luxe,” “literary girl.”
Dive Insight:
Etsy said the new shopping experience will allow for a deeper understanding of individual buyer behavior and its inventory. Over time, the company said it will have stronger data for machine learning to better predict buyer preferences.
The more a shopper engages with its platform the more personalized product recommendations become, according to Etsy.
Etsy's data from last year’s AI gifting recommendation tool showed a demand for more inspiration, which ultimately led to Wednesday's launch of AI-generated, curated collections, according to Daniel. The company is using a similar approach to create a more browseable experience.
“What we found was people really were craving more inspiration,” Daniel said in an interview with Retail Dive. “When we started with gift guides, we had a core set of gift guides that had a limited number of listings attached to them, and the resounding feedback that we got from our customers was they wanted more ideas. They also wanted more gifts per idea and that's when we started leveraging AI and LLM’s to both generate more ideas for us, but also to generate more of the gift ideas that were attached to each of those.”
In combining AI and human interaction, Etsy is using machine learning as a last step to make certain the company recommends all possible product options to a customer based on individual preferences.
“I think the real secret is leveraging LLM’s at the end to be able to then ask it to make sure there's cohesion across the collection,” Daniel said. Etsy asks AI to decipher if items fit in a given collection to “ensure that we haven't just algorithmically extended the seed collection, but we've done it in a way that still has that human touch.”
For Etsy, generative AI at the company has also improved attribute extraction, alt text for images, summarizing product reviews, enhancing accessibility and SEO, Daniel said.
While working to soothe artist-led AI anxiety, Etsy is incorporating AI across its platform, with precise delivery dates and the holiday gift recommendation tool.
Etsy's use of AI is “just scratching the surface right now,” Daniel said.
“AI has definitely been helping us push the product forward in ways that we would not have been able to otherwise,” Daniel said. “Both from a discovery perspective, we would not be able to be building these browsable, inspirational experiences, if not for AI, just because human curation does not scale enough to the extent that our customers would want. So if we're going to be that starting point for customers we're going to need to leverage AI as we are, but we are leveraging it across the board.”
