• For more than 20 years, Etsy has been the destination for human creativity online. • Our marketplace is home to more than 100 million special items made, handpicked and designed by more than 5 million sellers. • These items and the real people behind them are what set us apart. • But while the huge variety of Etsy’s inventory is one of our greatest strengths, it also creates fundamental engineering challenges specific to our marketplace. • The challenge: Etsy’s unique inventory With millions of creative items across thousands of categories - many of which are unique - it’s difficult to accurately capture all possible product attributes, which range from standard attributes like “color” and “material”, to niche attributes like “bead hole size” and “slime additives.” The range of possible attributes and their values is so broad that it’s a challenge even to enumerate them, let alone label listings with specific attribute data. • Unlike other online retailers (that may also have enormous inventories), because products on Etsy are listed by third party sellers and often handmade or customized, we do not have global SKUs (stock keeping units), or mappings from SKUs to product attributes.

Article Summaries:

  • Etsy’s marketplace hosts over 100 million handmade items from more than 5 million sellers, creating a vast inventory of unique products that lack standard SKUs or attribute mappings. While sellers can supply structured data (e.g., size, color), most listings rely on unstructured text and images, making it hard for search algorithms to quickly retrieve relevant results. Etsy’s engineering team has turned to large language models (LLMs) to automatically extract and convert this unstructured information into structured attributes, enabling faster filtering, comparison, and discovery across its diverse catalog. This initiative aims to improve buyer experience while maintaining low search latency.
  • Etsy, the online marketplace for handmade and unique items, faces a distinct engineering challenge: its inventory of over 100 million listings lacks standardized SKUs and often contains only unstructured seller data such as free‑text descriptions and photos. This makes it difficult to capture the full range of product attributes-both common (color, material) and niche (bead hole size, slime additives)-needed for effective search, filtering, and comparison. To address the latency and scalability issues of processing raw unstructured content, Etsy is developing large language models that convert free‑text and image data into structured attribute fields, enabling faster, more accurate discovery for buyers.
  • Etsy, the online marketplace for handmade and unique items, faces a major engineering challenge: its inventory of over 100 million listings lacks standardized product identifiers and often contains only unstructured seller data such as free‑text descriptions and photos. While structured attributes (size, color, material) enable efficient search and filtering, most sellers leave these fields blank to reduce friction, leaving critical details buried in text or images. To address this, Etsy is deploying large language models to automatically extract and convert unstructured information into structured attributes, thereby improving search relevance and discovery speed across its vast, diverse catalog.
  • Etsy’s 100 million‑plus listings, created by over 5 million sellers, lack a unified SKU system, making it hard to capture every product attribute-especially niche details such as bead hole size or custom materials. While sellers can supply structured data, most only provide free‑text titles, descriptions, and photos, leaving key information buried. Processing this unstructured content for every search query would hurt latency. To address this, Etsy is turning to large language models (LLMs) to automatically extract and convert unstructured text and image data into structured attributes, enabling faster, more accurate search filtering and comparison across its vast, unique inventory.
  • Etsy’s marketplace hosts over 100 million handmade items from more than 5 million sellers, but the sheer variety and lack of standard SKUs make it hard to capture every product attribute. Sellers often supply only free‑text titles, descriptions, and photos, leaving key details such as dimensions or material hidden in unstructured data. This hampers search and filtering, which rely on structured attributes. To address the latency and scalability challenges, Etsy is turning to large language models (LLMs) to automatically extract and standardize attribute information from listings, converting unstructured content into machine‑readable data that powers search, comparison, and discovery tools.

Sources: