Business organizations have limited advertising budgets. Consequently, Internet ad search engines should be designed to bid on advertisement auctions within budget constraints. The selection of ad auctions for competing advertisers is a two-sided problem. How does an auction search engine provide worthwhile ads to users, while simultaneously ensuring fail-safe investment profits for advertisers?
The authors of this paper investigate the issues of optimizing advertising budgets for auctions based on desired investment profits and ad quality. Typically, advertisers determine their participation in auctions within their budget constraints. However, the authors propose that a search engine could limit the participation of advertisers in auctions based on their budgets and ad objectives, such as increasing investment profits, decreasing cost-per-click, maximizing the quality of ads to display, and optimizing the ad quality relative to clicks.
To that end, the authors design and implement a system equipped with optimization algorithms for fair allocations of budgets based on ad objectives. In the algorithms, the historical ad traffic data and existing budgets are used to (a) compute the impression probabilities for allowing advertisers to participate in auctions; (b) rank the impressions of each advertiser based on the ad objectives, for use in selecting the best impressions until the budget is expended; and (c) compress the impression ranks into a histogram to decide if an advertiser with a budget constraint should participate in a given auction.
The accuracy and relevance of the bid optimized budget allocation algorithms for ad auctions are evaluated in offline and live simulation experiments. Contrary to the ad-auctioning algorithms proposed in this paper, at the exorbitant cost of processing times, the well-known linear programming algorithms [1,2] support the investigation of the effects of all budget constraints of advertisers. Unlike traditional bid algorithms [3], the authors offer an alternative bid-scaling algorithm to all advertisers.
This paper proposes a family of throttling algorithms for optimizing the limited ad search budget advertisements that promise future ad opportunities for businesses. Ideally, businesses would be able to make decisions by weighing clicks-per-dollar, profit-per-dollar, and multiple advertisement objectives using reliable optimization algorithms. The authors present reliable algorithms and results for accomplishing these goals. Theoretical scientists should weigh in on the tradeoffs between processing times and memory used by complex optimization algorithms that promote more accurate ad auctions within budget constraints.