Wayfair has deployed OpenAI models across catalog management and supplier support to reduce manual work, improve product discoverability and accelerate ticket handling. The company applied the models to validate attribute data at scale and to power intelligent routing and automated actions in its supplier support platform, producing measurable improvements in data quality and operational efficiency.
The retailer manages a catalog of roughly 40 million items, and accurate attribute tags—color, material, size and more—are critical to search, recommendations and merchandising. Maintaining those tags previously relied on supplier and customer reports and extensive manual review, a process that struggled to keep up with the catalog’s volume and complexity. To address this, Wayfair built a reusable catalog quality system that uses OpenAI models to generate guidance and then apply it to product records.
The system first assembles a label application guide from web information and Wayfair’s internal definitions, clarifying how attribute labels should be applied. The models then review product information against that guide and assign or correct labels. Wayfair reports the approach expanded the number of product details the AI can check at roughly 70 times last year’s pace. The AI-powered pipeline has processed more than 1 million products and validated and corrected attribute data for more than 11 million of the company’s most visible and frequently purchased items.
In controlled A/B testing, Wayfair found that products with corrected attribute data showed improvements in impressions, clicks and page rank, indicating better discoverability and increased shopper engagement. To manage risk and maintain supplier trust, the company uses a staged validation process: high-confidence corrections are applied automatically with supplier notification, while low-confidence or high-risk changes require supplier confirmation. Associates continue to sample outputs for audit, and suppliers participate in the validation loop. Jessica D’Arcy, Associate Director of Catalog Merchandising at Wayfair, noted that better data quality reduces downstream problems such as returns caused by inaccurate listings.
Wayfair also integrated OpenAI models into its supplier support platform, Wilma, to streamline the handling of tens of thousands of supplier interactions each month. Before the AI integration, associates routed incoming tickets manually across hundreds of issue types, a labor-intensive workflow that demanded broad institutional knowledge. Wilma now reads incoming requests, fills missing context using Wayfair’s databases, follows up with suppliers when necessary, and routes tickets to the appropriate internal teams.
The company said the integration moved from concept to production in under a month. Beyond routing, dozens of AI-driven workflows review historical records, assess complex cases, recommend next steps, and draft responses for human review. Wayfair measures how often AI recommendations match final human decisions through an alignment rate metric; when workflows meet predefined thresholds they can move from assistive co-pilot mode to semi-autonomous autopilot mode.
Overall, Wayfair reports automating 41,000 supplier support tickets per month, with automation reaching as much as 70% of ticket volume in some workflows. Teams report faster resolution times, higher supplier satisfaction, fewer reopened tickets and hundreds of saved hours previously spent on manual data entry and classification. Those operational gains have enabled broader issue coverage without requiring deep expertise across hundreds of topics, letting staff focus on higher-value interventions.
In addition to the custom integrations, Wayfair has rolled out more than 1,200 ChatGPT Enterprise seats across its roughly 12,000-person workforce. Employees use ChatGPT for routine tasks, internal problem solving and experimentation with generative models. The company also connected ChatGPT to services such as Slack and Google Drive to streamline daily workflows and make the models more useful across teams.
Looking ahead, Wayfair plans to extend the catalog validation capability to every new product added to its catalog, aiming to maximize visibility and sales from the start of each product’s lifecycle. The company continues to explore OpenAI’s latest models to tackle more ambiguous and contextual problems that traditional algorithms struggle to handle.
By combining catalog validation and supplier support automation, Wayfair has demonstrated how targeted use of OpenAI models can both improve customer-facing data and reduce internal manual effort. The company’s staged approach—balancing automatic corrections with supplier verification and human audits—offers a model for retailers that want to scale AI-driven quality improvements while managing risk and maintaining partner trust.
Source: Read the original source
Leave a Reply