Wayfair boosts catalog accuracy and support speed with OpenAI

Wayfair, the destination of all things home, helps shoppers find the right pieces for their homes from a catalog of roughly 40 million items. Keeping product information accurate and handling complex requests from tens of thousands of suppliers required substantial manual work. With OpenAI models, Wayfair built AI systems that validated over 11 million product specs and automated work across 41,000 supplier support tickets a month, with automation reaching up to 70% of tickets in some workflows. Together, these systems created more opportunities to grow sales from Wayfair’s existing catalog while freeing up team capacity by reducing manual work in catalog management and supplier support.

Wayfair’s catalog team manages tens of millions of products across nearly a thousand different product classes. Accurate tags for color, material, size, and other features power search, recommendations, and merchandising—helping shoppers more easily find pieces that fit their needs and feel confident about what they’re buying.

“The better our data quality, the more trust we build with the customer. It’s essential because it empowers shoppers to make the right buying decisions, directly reducing costly downstream issues like returns from misrepresented products,” said Jessica D’Arcy, Associate Director of Catalog Merchandising at Wayfair.

Wayfair evaluates how often the AI recommendations match the human agent’s final decision, a metric called “alignment rate.” Workflows that consistently reach a predetermined threshold can shift from assistive (“co-pilot”) to semi-autonomous (“autopilot”) modes. This staged evaluation approach helps teams decide when a workflow is ready for greater autonomy.

Before partnering with OpenAI, Wayfair’s tagging relied on suppliers and customers telling Wayfair that something looked wrong. Manual effort could not keep up with the volume. Early custom AI models for tagging were effective, but proved expensive at scale. With 47,000 tags to support, Wayfair needed a more effective approach.

Wayfair works with tens of thousands of suppliers each month. Wayfair associates previously reviewed each incoming supplier support ticket manually, identified what the supplier needed, and routed the issue to the appropriate team. This was a time-consuming and error-prone process. “Supplier support covers hundreds of different issue types,” said Brian Seaman, Head of Applied Science for Global Supplier at Wayfair. “It’s not realistic to expect any one associate to have deep expertise in every one of them.”

To help address these challenges, Wayfair added AI to help triage tickets in its custom supplier support platform called Wilma. With OpenAI models, the supplier support platform can now read incoming requests, fill in missing context, and route tickets to the right team. The existing OpenAI API integration helped the team move from idea to production in under one month. “Wilma gives associates the information they need to act,” said Seaman. “It reads the request, identifies what the supplier needs, brings in context from our databases, follows up with the supplier when needed, and routes the issue to the right team.”

Wayfair built an AI catalog quality system with OpenAI models to check and label product details, such as color, material, and size. First, the system uses web information and Wayfair’s internal definitions to create a guide explaining what each label means. The models then use that guide to review product information and assign the appropriate labels.

With this system, Wayfair is now expanding the range of product details its AI can check at 70 times last year’s pace. That helps Wayfair improve catalog accuracy with less manual work—making products easier to find and reducing the risk of returns caused by inaccurate listings.

“Our objective is to build trust so that customers are completely confident in what they are purchasing,” said Phillips. Wayfair developed a structured audit process in which associates inspect product samples to check the output, and suppliers help validate changes. When confidence is high, the system updates product content directly and notifies the supplier. Low-confidence or high-risk changes require supplier confirmation before taking effect.