ecommerce_ops · ecommerce · workflow
Wayfair builds LLM-powered style compatibility labeling pipeline on Google Cloud with Gemini 2.5 Pro
Wayfair's catalog of tens of millions of SKUs lacked scalable style-compatibility labels: traditional recommendation systems relied on behavioral signals that missed latent stylistic relationships, while human annotation was accurate but too slow and costly to cover the catalog at scale.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Pipeline pulls imagery and metadata
The batch pipeline pulls product imagery and metadata and builds structured prompts for the model.
Tools used
Gemini 2.5 ProGoogle Cloud
Outcome
The LLM-powered pipeline improved annotation accuracy by 11% over the initial generic prompt and generates style compatibility labels far faster than manual annotation, enabling scalable offline evaluation and future improvement of recommendation algorithms.
Results
Volume11%
Grounding & classification
Source type: technical build writeup
19 fields verified against source quotes.
computer visionpersonalizationrecommendation systemproduct catalogmetric backednamed customertools describedworkflow describedecommerceaccuracy improvementtime savedtechnical build writeupecommerce opsquality assurancedata sync enrichment