Timee's TEC algorithm boosts job-finding rate by 12.4% in field experiment

Timee, Japan's largest spot-work platform, tackles a classic recommender system challenge: when recommendations shape access to scarce, short-lived opportunities, maximizing predicted favoriting can lead to misdirected concentration. Popular job templates accumulate favorites but create few actual shifts, while templates with unmet labor demand get too little exposure. Researchers from Timee and academic partners designed exposure-control mechanisms for favorite-list management, reallocating template exposure based on posting activity and unfilled capacity.

The proposed algorithm, Thresholded Eligibility Control (TEC), is fully parallelizable and suitable for large-scale digital platforms. In simulations calibrated to Timee data, TEC increased the per-round job-finding rate from 57.6% to 70.0%—a 12.4 percentage point jump. A prefecture-level randomized field experiment confirmed real-world gains: more realized matches, higher exposure per active template, and a reduced share of low-exposure templates. The improvements extended to impression-level favoriting and downstream matching rates. The paper, published on arXiv under General Economics and Computer Science, demonstrates how thoughtful recommender design can balance user engagement with platform efficiency in gig economy settings.