LCM-Promo uses a finetuned LLM to allocate promotions
Original title: LCM-Promo: An LLM-Based Uplift Model for Promotion Allocation
In a churn-reactivation A/B test it raised conversion 1% and cut promotion cost 6% versus Segment-EMA.
Researchers present LCM-Promo, a task-specific Large Commerce Model finetuned on transactional data to predict user intent and promotion response. It reads single-token outputs as conversion-threshold probabilities and conditions on session context such as day and delivery shift. A production A/B test on churn reactivation beat the Segment-EMA baseline.
- The model learns monotonic promotion-response curves from production logs.
- Session context includes day and delivery shift.
- Conversion rose 1% and promotion cost fell 6% versus Segment-EMA.
From the original article
1 min read at arxiv.orgWe present LCM-Promo, an LLM-based uplift model for promotion allocation. LCM-Promo is a task-specific Large Commerce Model (LCM): a finetuned LLM trained on transactional commerce data to predict user intent and promotion response.
By Nesta Midavaine · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.