Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Automated Mechanism Design for Integrating Ad Auction and Allocation in Feed

Xuejian Li, Ze Wang, Bingqi Zhu, Fei He, Yongkang Wang, Xingxing Wang

TL;DR

This work addresses the problem of integrating ad auction and allocation in feed under externalities that affect CTR and GMV. It introduces MIAA, a deep automated mechanism with three modules—Externality-aware Prediction Module, Automated Auction Module, and Differentiable Sorting Module—that jointly predicts outcomes, selects the optimal allocation, and determines payments in an end-to-end fashion. The approach preserves incentive compatibility and individual rationality while maximizing revenue and GMV, and is validated through offline experiments on Avito and Meituan data and online A/B tests on Meituan, showing consistent improvements over strong baselines. The results demonstrate the practical viability of end-to-end mechanism design for feed ads and highlight the potential for extending to more complex multi-slot scenarios in real systems.

Abstract

E-commerce platforms usually present an ordered list, mixed with several organic items and an advertisement, in response to each user's page view request. This list, the outcome of ad auction and allocation processes, directly impacts the platform's ad revenue and gross merchandise volume (GMV). Specifically, the ad auction determines which ad is displayed and the corresponding payment, while the ad allocation decides the display positions of the advertisement and organic items. The prevalent methods of segregating the ad auction and allocation into two distinct stages face two problems: 1) Ad auction does not consider externalities, such as the influence of actual display position and context on ad Click-Through Rate (CTR); 2) The ad allocation, which utilizes the auction-winning ad's payment to determine the display position dynamically, fails to maintain incentive compatibility (IC) for the advertisement. For instance, in the auction stage employing the traditional Generalized Second Price (GSP) , even if the winning ad increases its bid, its payment remains unchanged. This implies that the advertisement cannot secure a better position and thus loses the opportunity to achieve higher utility in the subsequent ad allocation stage. Previous research often focused on one of the two stages, neglecting the two-stage problem, which may result in suboptimal outcomes...

Deep Automated Mechanism Design for Integrating Ad Auction and Allocation in Feed

TL;DR

This work addresses the problem of integrating ad auction and allocation in feed under externalities that affect CTR and GMV. It introduces MIAA, a deep automated mechanism with three modules—Externality-aware Prediction Module, Automated Auction Module, and Differentiable Sorting Module—that jointly predicts outcomes, selects the optimal allocation, and determines payments in an end-to-end fashion. The approach preserves incentive compatibility and individual rationality while maximizing revenue and GMV, and is validated through offline experiments on Avito and Meituan data and online A/B tests on Meituan, showing consistent improvements over strong baselines. The results demonstrate the practical viability of end-to-end mechanism design for feed ads and highlight the potential for extending to more complex multi-slot scenarios in real systems.

Abstract

E-commerce platforms usually present an ordered list, mixed with several organic items and an advertisement, in response to each user's page view request. This list, the outcome of ad auction and allocation processes, directly impacts the platform's ad revenue and gross merchandise volume (GMV). Specifically, the ad auction determines which ad is displayed and the corresponding payment, while the ad allocation decides the display positions of the advertisement and organic items. The prevalent methods of segregating the ad auction and allocation into two distinct stages face two problems: 1) Ad auction does not consider externalities, such as the influence of actual display position and context on ad Click-Through Rate (CTR); 2) The ad allocation, which utilizes the auction-winning ad's payment to determine the display position dynamically, fails to maintain incentive compatibility (IC) for the advertisement. For instance, in the auction stage employing the traditional Generalized Second Price (GSP) , even if the winning ad increases its bid, its payment remains unchanged. This implies that the advertisement cannot secure a better position and thus loses the opportunity to achieve higher utility in the subsequent ad allocation stage. Previous research often focused on one of the two stages, neglecting the two-stage problem, which may result in suboptimal outcomes...
Paper Structure (21 sections, 19 equations, 2 figures, 6 tables)

This paper contains 21 sections, 19 equations, 2 figures, 6 tables.

Table of Contents

  1. Introduction
  2. related work
  3. Preliminaries
  4. Setting for Ad Auction and Allocation
  5. Problem Formulation
  6. Automated Mechanism Design
  7. Methodology
  8. Externality-aware Prediction Module
  9. Automated Auction Module
  10. Differentiable Sorting Module
  11. Training and Online Serving
  12. Experiments
  13. Experimental Settings
  14. Dataset
  15. Evaluation Metrics
  16. ...and 6 more sections

Figures (2)

  • Figure 1: On Meituan retail delivery platform, the ad 'LAYS CHIPS' in the feed is presented to user along with external organic items. Whether a user clicks on the ad is easily impacted by both the position and the context of the ad.
  • Figure 2: The architecture of MIAA. MIAA consists of three modules: Externality-aware Prediction Module (EPM), Automated Auction Module (AAM) and Differentiable Sorting Module (DSM).