Pricing Guidance in Ad Sale Negotiations:The PrintAds ExampleAdam Isaac [email protected]. [email protected] [email protected] consider negotiations between publishers and advertis-ers in a marketplace for ads. Motivated by Google’s onlinePrintAds system which is such a marketplace, we focus onthe role of the market runner in improving market efficiency.We abstract the problem of pricing guidance where the mar-ket runner provides an initial price-point for negotiationsbased on data analysis. The problem is nuanced because themarket runner can not fully reveal the price data for any ofthe publishers. We introduce two solutions for pricing guid-ance, the first using clustering and the second using supportvector machines, and present experimental evaluation of ourmethods. Pricing guidance by the market runner is a noveldirection, and we expect more research in the future.1. INTRODUCTIONPrices for goods and services are determined in differentways. Traditionally, prices are set by the seller or provider,such as in retail stores (per item) or telephone bills (per callor as a flat fee per month). A different way to set prices isvia an auction, and this is increasingly used on the Inter-net, including in ad sales. The focus of this paper is on yetanother way prices are determined, via negotiation, which isused extensively in marketplaces. In particular, we consideronline marketplaces where parties negotiate over the Web.The motivating example for us is that of Google’s AdSensefor Print product (heretofore referred as PrintAds) [1] - amarketplace for newspaper publishers and advertisers to ne-gotiate prices for ads that appear in print. In what follows,we will provide an overview of PrintAds, motivate the prob-lem of providing pricing guidance to the parties, and offer asolution to this problem.PrintAds. PrintAds is a marketplace for print publishersand advertisers. Advertisers log into the PrintAds systemand are helped by various targeting tools to identify desir-able publishers (e.g., finding a New York metro daily news-paper with circulation above 10,000 readers). Then, theyenter the details of the ad they wish to publish, such as itsADKDD’09, June 28, 2009, Paris, France.size, the desired section where they wish their ad to appearetc. Finally, advertisers specify an offer price they are will-ing to pay for the ad to run. When they submit an offer, itis sent to the publisher, who may accept the offer or decline.If the publisher accepts, the ad eventually runs and paymentrendered; else, a new round of negotiation begins.1The two primary parties within this marketplace are thepublishers and the advertisers. Publishers partake in thismarket in order to reach a larger scale of potential adver-tisers than would otherwise be possible through their di-rect sale channels, while advertisers partake in the marketas it more readily enables them to negotiate with multiplepublishers easily. However, an important third participantwithin this market is Google as the market runner. Themarket runner ensures negotiations proceed smoothly andseeks to make the market as efficient as possible. Efficiencyis desired by the market runner not only for altruistic rea-sons, but also as the market runner receives compensationonly when transactions are successfully completed, and sohas economic incentives to increase overall value creation.Pricing in PrintAds. Traditionally ads in print have beensold via direct sales teams. While rate cards exist withposted prices, negotiations typically occur between the indi-vidual advertisers and the publisher, resulting in contractsthat vary widely in price, different from the rate card maxi-mums. These characteristics of contracts can also be presentwithin PrintAds. However, because the typical PrintAds ad-vertiser is new to advertising within newspapers,2negotia-tions tend to proceed less smoothly than offline negotiations:some offers made by advertisers may be viewed as exceed-ingly low by publishers, some publishers do not respond tooffers tendered, etc. In order to smooth the negotiation pro-cess, the market runner needs to provide better guidanceabout typical prices of print inventory. This is a nuancedrole. Publishers seek higher prices, close to their rate cards,but also do not want to forego the incremental income thatthis additional sales channel can provide. Advertisers likelower prices, but want to secure that price without having tobear the costs of entering a negotiation with the direct sales1The system and its variants are described in more detail inSection 2.2Many PrintAds customers are more familiar with Google’sAdWords program, where they construct small, text-onlyadvertisements, and pay very small amounts of money ona per click basis. This is in contrast to advertising in anewspaper where creatives are richer and payments tend tobe significantly higher.channel offline. For both parties to operate under similarexpectations around a reasonable set of prices, the marketrunner needs to provide pricing guidance; however, the mar-ket runner does not get to set prices as these are ultimatelydetermined by negotiations between the two parties. Thegoal of the market runner is only to help negotiations reachan agreement. Additionally, the market runner has to becareful that it does not compromise any private informationof the publisher or advertiser during negotiation. This in-troduces a game-theoretic perspective to the design of suchguidance.Our Contributions. We approach the problem of prov-ing pricing guidance as one of data analysis. As an analogy,consider how buyers make offers on real estate. There istremendous strategy in how sellers arrive at the list priceby considering comparable sale data, both to target a sig-nificant group of buyers as well as to spur a bidding war.Likewise, in a buyers’ market, the buyers have an eventualprice they are willing to pay, but strategize while makingrounds of offers. The PrintAds marketplace offers far morecomplexities over this analogy. Publishers’ inventory of adspace is flexible, but market demand at times can be thin,and online negotiations can run to many rounds. Our con-tributions are as follows.• We model the Go ogle PrintAds marketplace and ab-stract the pricing guidance problem. In particular,we address a concern that price prediction should notoverfit any particular publisher’s data.• We propose two algorithms based on data analysis tosolve the pricing