Disaggregate Attribution Analysis on Web Advertising

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K03972
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Commerce
• Research Institution: Gunma University (2018); Nagasaki University (2016-2017)
• Principal Investigator: Takahashi Kei 群馬大学, 数理データ科学教育研究センター, 准教授 (70595280)
• Research Collaborator: Hirai Hirohide
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 広告効果 / 心理的効果 / ベイズ推定

Outline of Final Research Achievements

This study develops a hidden Markov model that has hidden and discrete states of consumers' attitude for advertisement. In this model, we consider three psychological effects on advertisement, forgetting, wear-in and wear-out effects. We apply this model for data at the EC site. As a result of analysis, sign conditions on parameters in psychological effects. Additionally, we reveal that there exists little wear-in effects in this EC site data. Secondly, we conduct attribution analysis based on our model. Consequently, we get different attribute allocation from that without considering psychological effects.

Free Research Field

マーケティング・サイエンス

Academic Significance and Societal Importance of the Research Achievements

本研究では，Webサイト等へのアクセス履歴データにより，従来集計的な把握に留まっていた心理的効果，具体的には忘却効果，刷込み効果，擦切れ効果について非集計的に存在を把握することができた．また，Web広告において多用されていたアフィリエイトやリターゲティング広告について，その広告を"踏んだ"ことによる離脱を考慮すると，最終目標への貢献度が低いことが分かった．