Product descriptions:
Furniture.com is transforming online furniture shopping by unifying a fragmented industry with data and AI. Partnering with 60+ retailers and offering over 1.5 million SKUs, the company brings a vast catalog into one seamless platform. One of their flagship innovations is Find It AI, an AI-powered product discovery tool built into their e-commerce website. It powers the shopping experience by using image-based search to deliver more accurate results, enabling users to find their perfect piece faster. Built on the Databricks Data Intelligence Platform, Furniture.com leverages Delta Lake for data processing at scale, MLflow for model experimentation and Unity Catalog for granular access control. These tools streamline data workflows, automate processes and allow for dynamic, personalized collections that create a truly intuitive shopping journey, driving higher engagement, retention and sales.
Helping shoppers find the perfect piece of furniture
Furniture.com is transforming the online furniture shopping experience by bringing together over 60 retail partners into one unified shopping platform, creating a seamless and personalized shopping journey for customers. However, the fragmented nature of the furniture industry posed significant challenges, from inconsistent image sizes and missing attributes to unstructured data across over 1.5 million SKUs.
“There are no standards in the furniture industry,” Daniel Russotto, General Manager of Furniture.com, said. “Image sizes are different, and every single piece of data for each retailer is different. When we first started, every time new languages came in, we had to map it again.”
The company’s mission to simplify the shopping journey required a solution that could unify this data and deliver the right responses to customer searches. Still, manual processes for cleaning and standardizing product information quickly became impractical. “If we standardized a sofa color from lilac to purple for one partner, we wouldn’t be able to reuse that logic elsewhere,” Daniel explained.
Building an AI-powered search tool that understands the customer
With the Databricks Data Intelligence Platform, Furniture.com now runs their entire ML lifecycle — from ingestion and transformation to model training and deployment — on a single platform. “Everything happens in Databricks,” Yasel Garces, Director of Data Science and ML at Furniture.com, said. “It’s very easy for us to consume the data, train our models, test, deploy and monitor.”
“Every time the user goes to our website and performs a search, they have an AI experience, whether they know it or not. Behind the scenes, it’s connecting their ideas with real products,” Yasel said. “Under the hood, we have a model that takes the query as an input, generating a synthetic image using generative AI, and then we are using similar items. We compare that image with our catalog and come back with recommended products that align with the user vision.”
Furniture.com’s Find It AI uses generative AI to generate images that represent shopper intent. Then it matches the image with their product catalog, delivering personalized, relevant search results.
To improve product visuals, the team also built computer vision models for image classification and cropping. “We have a series of computer vision models to standardize them into one clean, sleek image,” Brianna King, Lead Data Scientist at Furniture.com, said. “The classification model enables us to identify a lifestyle image — a more visually appealing image.” Another production model is called Collections, which uses large language models (LLMs) to group related products, like bedroom or dining sets, even when partners don’t explicitly label them. “We were able to find more than 16,000 collections across our 50+ partners,” Yasel said. “This model is fully integrated into our ETL process and runs automatically — no human intervention.”
The team also uses Databricks Apps to share interactive QA tools across the company. “Instead of having two QAs, you can have a whole company available to easily QA the model,” Yasel said. “It helps everybody in the company to understand the product better — but also it’s a way to get feedback before we put this in production and show it to the world.”
Driving higher customer engagement and conversion
Since launching their AI-powered features on Databricks, Furniture.com has seen dramatic improvements in user engagement and conversion. “Since deployment, we’re seeing a very positive impact in how users interact with our site and return to it,” Yasel said.
The average click-through rate (CTR) across Furniture.com is 3%. However, users who see the Find It AI experience have a CTR of 10%, and those who actively interact with it show a CTR of 23.5%, representing more than an 8x increase from the baseline. The impact is just as strong for repeat visitors. The baseline return rate for Furniture.com is 7.17%. For those who interact with the Find It AI experience, the return rate climbs even higher to 23.2%, representing a 3.2x improvement.
Operationally, the company no longer needs to manually standardize attributes or categorize products — those processes are now automated with ML models embedded in the platform. This has allowed the team to onboard new partners faster while freeing up internal resources. Daniel noted that the merchandising team operates leanly: just four or five people supporting a catalog of over 1.5 million SKUs. Without automation from Databricks, he estimated, “that team would need to be 10+.”
Looking ahead, the team continues to invest in AI innovation — from lifestyle image generation to semantic search — all built and deployed on Databricks.