ecommerce_ops · ecommerce · workflow
Target's GRAM model uses LLMs to boost Home accessory recommendation conversions by 12%
Providing high-quality accessory recommendations across Target's vast catalog was infeasible manually because of the sheer volume of items and the many product attributes that must be considered for each pairing.
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 · LLM attribute importance weighting
LLMs automatically analyze product data to identify the most important attributes and assign them importance weights for pairs of core and accessory item types.
Tools used
LLMsGRAM
Outcome
A/B testing of the Home Accessory model in the add-to-cart flyout in February 2025 showed around an 11% increase in interaction rate, roughly a 12% increase in display-to-conversion rates, and more than 9% growth in attributable demand; the model was fully rolled out to production in April 2025.
Results
Volumearound 11% increase
Running sinceApril 2025
Grounding & classification
Source type: technical build writeup
22 fields verified against source quotes.
data extractionrecommendation systemproduct cataloghuman review describedmetric backednamed customerproduction runtime claimedtools describedworkflow describedretailconversion increasecustomer satisfactionrevenue increasetechnical build writeupecommerce opsai draft human approvalextract classify route