On the Benefits of Keyword Spreading in Sponsored Search Auctions: An Experimental Analysis

Sellers of goods or services wishing to participate in sponsored search auctions (SSA) must define a pool of keywords that are matched on-line to the queries submitted by the users to a search engine. Sellers must also define the value of their bid to the search engine for showing their advertisements in case of a query-keyword match. In order to optimize its revenue a seller might decide to substitute a keyword with a high cost, thus likely to be the object of intense competition, with sets of related keywords that collectively have lower cost while capturing an equivalent volume of user clicks. This technique is called keyword spreading and has recently attracted the attention of several researchers in the area of sponsored search auctions. In this paper we describe an experimental benchmark that, through large scale realistic simulations, allows us to pin-point the potential benefits/drawbacks of keyword spreading for the players using this technique, for those not using it, and for the search engine itself. Experimental results reveal that keyword spreading is generally convenient (or non-damaging) to all parties involved.