ecommerce_ops · ecommerce · workflow
leboncoin deploys LLM-powered Re-Ranker to improve search relevance and lift click and contact rates
leboncoin faces a complex search challenge: nearly 30 million monthly users and over 60 million classified ads each described in free-form user language create high volatility, while the search engine must respond within tens of milliseconds at thousands of requests per second, making relevant ranking technically demanding.
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 · Ad embeddings pre-computed
The Ad Encoder computes vector representations of all catalogue ads and stores them in a vectors database ahead of real-time serving.
Tools used
ElasticSearchLLMvectors database
Outcome
The Re-Ranker improved business targets (click and contact rate) by up to 5% and user experience KPIs (nDCG and average clicked and contacted positions) by up to 10%.
Results
Volumeup to +5%
Grounding & classification
Source type: technical build writeup
19 fields verified against source quotes, 1 dropped as unverifiable.
predictive analyticsrecommendation systemproduct catalogmetric backednamed customerproduction runtime claimedsource backedworkflow describedecommerceaccuracy improvementconversion increasecustomer satisfactiontechnical build writeupecommerce opsextract classify route