User Response Learning for Directly Optimizing Campaign Performance in Display Advertising

Abstract

Learning and predicting user responses, such as clicks and conversions, are crucial for many Internet-based businesses including web search, e-commerce, and online advertising. Typically, a user response model is established by optimizing the prediction accuracy, e.g., minimizing the error between the prediction and the ground truth user response. However, in many practical cases, predicting user responses is only part of a rather larger predictive or optimization task, where on one hand, the accuracy of a user response prediction determines the final (expected) utility to be optimized, but on the other hand, its learning may also be influenced from the follow-up stochastic process. It is, thus, of great interest to optimize the entire process as a whole rather than treat them independently or sequentially. In this paper, we take real-time display advertising as an example, where the predicted user's ad click-through rate (CTR) is employed to calculate a bid for an ad impression in the second price auction. We reformulate a common logistic regression CTR model by putting it back into its subsequent bidding context: rather than minimizing the prediction error, the model parameters are learned directly by optimizing campaign profit. The gradient update resulted from our formulations naturally fine-tunes the cases where the market competition is high, leading to a more costeffective bidding. Our experiments demonstrate that, while maintaining comparable CTR prediction accuracy, our proposed user response learning leads to campaign profit gains as much as 78.2% for offline test and 25.5% for online A/B test over strong baselines.

Display Advertising with Real-Time Bidding (RTB) and Behavioural Targeting

“…It could be more crucial to take the market information into consideration in the utility function for second-price auctions. Finally, one may take a step forward to coordinate the optimisation for both CTR estimator and bidding strategy to further push the bidding performance [Ren et al, 2016]. [Zhang and Wang, 2015] studied the problems of arbitrages in real-time bidding based display advertising in a multi-campaign setting.…”

Bidding Machine: Learning to Bid for Directly Optimizing Profits in Display Advertising

“…However, almost all the related work bases bidding decision on the estimated utility and the predicted cost, and treats these three parts desperately. [33] embedded user response prediction, i.e., utility estimation, into the bid optimization and made the prediction aware of the market information and cost sensitivity. [48] took budget constraint, winning probability and cost of the particular ad impression altogether and proposed a functional bidding optimization to maximize the target KPIs, e.g., clicks.…”

“…Following [7], [33], the overall revenue R(·) can be defined as utility function below w.r.t. the corresponding auction sample (x, y):…”

“…According to the Bayesian decision theory [4], the learning of the user response model and bid landscape model should be informed by the final bidding utility. As is studied in the literature [33], the required accuracy of the CTR prediction would not be the same throughout the range of the prediction [0, 1] as there is a cost (negative utility) for the advertiser to win an impression if no click, but no cost (zero utility) for losing one. The value of clicks also varies across campaigns; and it would be good if the CTR learning can tailor its efforts more toward those higher-valued cases and make them better predicted.…”

“…To our best knowledge, it is the first work that takes these three key components of RTB altogether to optimize. In our recent work [33], we combined utility estimation and bidding strategy in a whole while adopting naive counting-based method for market price modeling, which is not optimal considering the expected cost since the market price is flexible in different contexts. The authors in [22] proposed a method which combines CTR estimation and winning price prediction (as will be compared in the experiment).…”