Slide 1Slide 2Slide 3Slide 4Slide 5Slide 6Slide 7Slide 8Slide 9Slide 10Slide 11Slide 12Slide 13Slide 14Slide 15Slide 16Slide 17Slide 18Slide 19Slide 20Slide 21Slide 22Slide 23Slide 24Slide 25Slide 26Slide 27Slide 28Slide 29Slide 30Slide 31Slide 32Slide 33Slide 34Slide 35Slide 36Slide 37Slide 38Slide 39Slide 40Slide 41Slide 42Slide 43Online Advertising and Ad Auctions at GoogleVahab MirrokniGoogle Research, New YorkTraditional AdvertisingAt the beginning: Traditional AdsPosters, Magazines, Newspapers, Billboards.What is being Sold:Pay-per-Impression: Price depends on how many people your ad is shown to (whether or not they look at it)Pricing:Complicated Negotiations (with high monthly premiums...)Form a barrier to entry for small advertisersAdvertising On the WebOnline Ads:Banner Ads, Sponsored Search Ads, Pay-per-Sale ads.Targeting:Show to particular set of viewers. Measurement: Accurate Metrics: Clicks, Tracked Purchases.What is being Sold:  Pay-per-Click, Pay-per-Action, Pay-per-ImpressionPricing:AuctionsHistory of Online Advertising1994: Banner ads, pay-per-impressionBanner ads for Zima and AT&T appear on hotwired.com.1998: Sponsored search, pay-per-click 1st-price auctionGoTo.com develops keyword-based advertising with pay-per-click sales.2002: Sponsored search, pay-per-click 2nd-price auctionGoogle introduces AdWords, a second-price keyword auction with a number of innovations.1996: Affiliate marketing, pay-per-acquisitionAmazon/EPage/CDNow pay hosts for sales generated through ads on their sites.Banner AdsPay-Per-ImpressionPay-per-1000 impressions (PPM): advertiser pays each time ad is displayedModels existing standards from magazine, radio, televisionMain business model for banner ads to dateCorresponds to inventory host sellsExposes advertiser to risk of fluctuations in marketBanner blindness: effectiveness drops with user experienceBarrier to entry for small advertisersContracts negotiated on a case-by-case basis with large minimums (typically, a few thousand dollars per month)Sponsored Search AdsPay-Per-ClickPay-per-click (PPC): advertiser pays only when user clicks on adCommon in search advertisingMiddle ground between PPM and PPADoes not require host to trust advertiserProvides incentives for host to improve ad displaysAuction MechanismAdvertisements sold automatically through auctions: advertisers submit bids indicating value for clicks on particular keywordsLow barrier-to-entryIncreased transparency of mechanismKeyword bidding allowed increased targeting opportunitiesAuction MechanismInitial GoTo model: first-price auctionAdvertisers displayed in order of decreasing bidsUpon a click, advertiser is charged a price equal to his bidUsed first by Overture/Yahoo!Google model: stylized second-price auctionAdvertisers ranked according to bid and click-through-rate (CTR), or probability user clicks on ad Upon a click, advertiser is charged minimum amount required to maintain position in rankingBidding PatternsGraph from [Zhang 2006]Bidding history in Yahoo! First-Price Auction:Bidding PatternsGraph from [Zhang 2006]Bidding Process4Targeting PopulationsAdvert CreationKeyword SelectionBids and Budget 321“You don’t get it, Daddy, because they’re not targeting you.”Bidding Process4Targeting PopulationsAdvert CreationKeyword SelectionBids and Budget “Here it is – the plain unvarnished truth. Varnish it.”321Ad titleAd textDisplay urlBidding Process4Targeting PopulationsAdvert CreationKeyword SelectionBids and Budget “Now, that’s product placement!”321Bidding Process4Targeting PopulationsAdvert CreationKeyword SelectionBids and Budget 321Daily BudgetAuction MechanismA repeated mechanism!Upon each search, Interested advertisers are selected from database using keyword matching algorithmBudget allocation algorithm retains interested advertisers with sufficient budget Advertisers compete for ad slots in allocation mechanismUpon click, advertiser charged with pricing schemeCTR updated according to CTR learning algorithm for future auctionsClick-Through RatesClick-through rate (CTR): a parameter estimating the probability that a user clicks on an adA separate parameter for each ad/keyword pairAssumption: CTR of an ad in a slot is equal to the CTR of the ad in slot 1 times a scaling parameter which depends only on the slot and not the adCTR learning algorithm uses a weighted averaging of past performance of ad to estimate CTR(Old) Yahoo! 2nd-Price AuctionAdvertiserABCBid Allocation Price$10 2 $5$5$50X1$0$10per click!Ad slot 1Ad slot 2KeywordAlgorithmic search resultsGoogle Single-Shot AuctionAdvertiserABCBid CTR Bid x CTR Allocation Price$10 0.10 1.0 2 $5$5$500.500.012.50.51X$2$0(expected bid per impression)per click!Ad slot 1Ad slot 2KeywordAlgorithmic search resultsKeyword MatchingExact match: keyword phrase equals search phrase Phrase match: keyword phrase appears in search (“red roses” matches to “red roses for valentines”)Broad match: each word of keyword phrase appears in search (“red roses” matches to “red and white roses”)Issues:Tradeoff between relevance and competitionHow to handle spelling mistakesBudget AllocationBasic algorithmSpread monthly budget evenly over each dayIf budget leftover at end of day, allocate to next dayWhen advertiser runs out of budget, eliminate from auctionsIssues:Need to smooth allocation through-out dayAllocation of budget across keywordsTypical ParametersKeyword Price in 3rd slot# of Keywords $20-$50 2$10.00 - $19.99 22 $5.00 - $9.99 206 $3.00 - $4.99 635 $1.00 - $2.99 3,566 $0.50 - $0.99 4,946 $0.25 - $0.49 5,501 $0.11 - $0.24 5,269 PPC of most popular searches in Google, 4/06Typical ParametersKeyword Top Bid 2nd Bidmesothelioma $100 $100structured settlement $100 $52vioxx attorney $38 $38student loan consolidation $29 $9Bids on some valuable keywordsCTRs are typically around 1%Other Important issues in ad auctionsAvoiding click fraudBidding with budget constraintsExternalities between advertisersUser search modelsMeasurement: Information Adwords FrontEnd: Bid SimulationsClicks and Cost for other bids. Google AnalyticsTraffic Patterns, Site visitors. Search insights:Search Patterns Interest-Based AdvertisingIndicate your interests so that you get more relevant