Less Traffic, Better Outcomes: Competition-Aware Request Dispatch in Real-Time Ad Exchanges

2026-08-04

AI Summary

This work addresses the inefficiency in real-time advertising auctions caused by excessive request distribution, which reduces demand-side platform (DSP) participation due to computational and budget constraints, thereby impairing monetization. The authors propose the first competition-aware request dispatching framework that dynamically determines whether to forward each ad request to individual DSPs through distributed bid prediction and a probabilistic forwarding mechanism. Coupled with a lightweight online policy optimizer, the framework adaptively adjusts decision thresholds under non-stationary market conditions. Without increasing total request volume, the approach intelligently reveals comparative advantages among DSPs. Deployed in a production system handling over 20 billion daily requests, it reduced DSP request load by 34.2% and significantly increased net revenue by 4.6% (p<0.001), with heterogeneous effects observed across traffic segments.

Abstract

Real-time bidding (RTB) ad exchanges typically forward nearly all incoming requests to demand-side platforms (DSPs), even though only a small fraction receive bids. This over-distribution weakens auction outcomes: DSPs throttle participation under compute and budget constraints, reducing the effective use of limited bidding capacity. We present a competition-aware request dispatch framework that uses distributional bid prediction and probabilistic forwarding to decide whether each request should be sent to each DSP. The system adapts per-DSP thresholds over time through lightweight policy optimization to track non-stationary market conditions. We evaluate the framework through four sequential online experiments on a production platform serving over 20 billion daily requests. A full multi-DSP deployment reduces DSP request volume under the policy by 34.2% while increasing net revenue by 4.6% (p<0.001) in a recent 14-day window after an initial DSP adaptation period. Further analysis highlights strong heterogeneity across traffic segments and reveals that aggregate metrics can be misleading. Segment-level and per-DSP analyses suggest that the policy surfaces comparative advantages among DSPs, improving monetized outcomes without increasing overall request volume.

Problem

• real-time bidding
• ad exchange
• request dispatch
• auction efficiency
• demand-side platform

Innovation

• competition-aware dispatch
• distributional bid prediction
• probabilistic forwarding
• real-time bidding
• policy optimization