Changhao Jiang, Min Zhang, Bin Gao, and Tie-Yan Liu
This paper is a study on the advertisers from whom the search engine may increase the revenue by offering an advanced sponsored search service. To better understand the advertisers in sponsored search, we divide them into head advertisers and tail advertisers according to their contributions to the search engine revenue. Data analysis shows that though some tail advertisers are with bad performances, they actually have large amount of budgets and their budget usage ratio is quite low. That is, these advertisers have potential to be large revenue contributors. They aimed to achieve the planned campaign goals (e.g., a large number of clicks), but they finally failed in doing so due to wrongly-selected bid keywords, inappropriate bid prices, and/or low-quality ad creatives. In this paper, we conduct a deep analysis on these advertisers. Specially, we define the measures to distinguish potential head advertisers from tail advertisers, and then run simulation experiments on the potential head advertisers by applying different improving methodologies. The results show that the performances of the potential head advertisers are easy to be improved if we correctly identify the primary cause for the underperformance. We also show that a decision tree model can be implemented for the bad performance diagnosis.
In Lecture Notes in Computer ScienceVolume 7675, 2012, pp 174-186