We use cookies. You have options. Cookies help us keep the site running smoothly and inform some of our advertising, but if you’d like to make adjustments, you can visit our Cookie Notice page for more information.
We’d like to use cookies on your device. Cookies help us keep the site running smoothly and inform some of our advertising, but how we use them is entirely up to you. Accept our recommended settings or customise them to your wishes.

Google's Revenue Dials

My monthly column for Search Engine Land, in case you missed it: Occasionally I'm asked by investment bank types whether Google can "dial up" their revenue on a whim, and moreover whether they do so. The short answer is "certainly, yes" to the first part and "probably no" to the second. It's worth spending a bit of time to detail Google's three major revenue dials. {Note: Most of these arguments apply equally well to Bing, but let's stick to the vernacular of Google for simplicity.} Dial #1: How Much Weight to Give "Relevance" The most powerful control in Google's arsenal is its ability to select which ads are relevant to a given user search. On the SERP ads from different advertisers will appear in order of their Ad Rank based on the following formula: Ad Rank = Max CPC Bid x Quality Score Quality Score is further defined as a measure of anticipated CTR, keyword relevance, and landing page quality. Landing page quality does not factor into the ad's position on the SERP, but does impact the CPC paid. So really it's just 3 factors: Bid, anticipated CTR, and relevance. Google maximizes its immediate term revenue by ignoring relevance all together. Consider the following example: User Search = "furniture" The following KW are available to Google with the associated estimates for first position CTR and bid, my own assessment of their relevancy, and the expect Revenue Per Thousand Impressions Pretty clearly, Mesothelioma would win if Google ignored relevancy entirely and focused strictly on what made them the most money on that query. Now, the law firm or lead aggregator running that ad may not want that ad to be served quite so broadly, but Google gets to define how broad is broad, and while this example is exaggerated to make a point, pretty clearly how much weight relevancy gets will impact the ordering of the ads and will do so in a way that impacts Google's revenue. The same basic process takes place in choosing which of a given advertiser's ads are served. For advertisers using any phrase- or broad-matched ads, Google has a choice of which ad to serve, and this flexibility also allows Google to dial up or dial down its immediate term revenue. A useful way to think about it is that immediately prior to the grand auction is a mini-auction taking place within each advertiser's account. The distinction being that for this "Pre-Auction" only the winner advances to the grand auction, there is no second place. Google describes this selection process as working through an ordered set of preferences, with exact matches of the user query taking the highest preference (over Ad Rank). However, this preference can be overridden by other factors (like Ad Rank) if the discrepancy is high, and we find that the stated preference for the exact keyword match seems to disappear if that exact matched keyword is not bid to the first page minimum. As with the main auction, the more weight Google places on relevance the less money they make in the short term. Similarly, Google's insistence on some level of relevancy and commercial intent is evidenced by search phrases that return no ads whatsoever. Take for example the user search "F-Stop" Undoubtedly, photography shops would like to serve ads to these folks, but Google suppresses these ads because the CTR is so low that Google has determined that these users seek information, not products and services. Google chooses to make no money from these searches rather than a little bit. Clearly there is a balance being struck between long-term interests and short-term interests. Google knows that if it puts too little weight on relevance users will eventually tune out the ads as visual noise and having had bad experiences clicking on irrelevant ads, they'll stop using them. Moreover, advertisers will reduce bids as the quality of traffic deteriorates from serving the wrong ad to the wrong person. However, Google controls that balance. Dial #2: Real Estate Allocation How much space is devoted to ads versus organic listings? Google increased its RPM materially last year by moving the ads on the right rail a few pixels towards the center. Could they move them a few pixels more? Or, increase the font size/darkness of the ads to make them stand out more? Google has also:
  • made the sponsor listings background a shade or two closer to white and forced display urls to be all lower case. The net effect of this has been to make ads appear more like the organic listings. Perhaps not the intent, but certainly the effect.
  • they've added sitelinks which both take up more real estate and ...make the ads look more like organic links...
  • and, Google changed the wording from "sponsored listings" to "ads". Perhaps to more explicitly comply with FTC regulations...or to call less attention to them by reducing the character count...
What about that notion of commercial intent? If they've taken the position that when commercial intent is low they should only serve organic listings, couldn't they take the opposite view when commercial intent is high? Currently 3 is the greatest number of ads you see in "promoted position" above the organic listings, but why couldn't that be 4, 5 or 10 if the commercial intent suggests that's what users want? Dial #3: Threshold Levels Last, and somewhat more subtle. Google sets thresholds for appearing on the first page and for appearing in "promoted position" above the organic listings. As Mark Ballard and Matt Mierzejewski described over at RKG Blog, this has the net effect of increasing the CPCs paid by the last advertiser above that threshold. Google kind of acts like another advertiser bidding in the landscape. We don't know whether the threshold Ad Rank values change in response to that last advertiser's Ad Rank. Certainly Google could effectively ratchet up the threshold value to maximize the last advertiser's CPC paid, but it's doubtful they do so. Conclusions The existence of these revenue dials doesn't imply their use, necessarily. I did a bit of research to see if I could find evidence that Google amped up the short term revenue dials prior to quarterly earnings calls but could find no such evidence in our data. Anyone in the industry who's been paying attention would see that they do seem to play with the nobs, but we can't detect any patterns to it, and much of this is opaque enough to be difficult to detect. Moreover, if Google did jigger with the dials with specific intent to dial up revenue I'm not sure we'd have much to complain about. In no case do we pay more than we bid, so if Google makes adjustments that impact our ROI we have the tools to keep our economics in balance. Most businesses have the ability to amp up their margins in the short run, so this isn't anything new, really. The "old-timers" of the industry may remember that Overture offered advertisers a choice of paying exactly what you bid or paying a penny more than the person below you. Now, who in their right mind would choose the former? Probably no one would make that choice, but we took over many an account that did just that. Google could do the same and be perfectly within their rights as the publisher (indexer?). I hope they don't, but the point is it's up to the managers of paid search to react to the game as it evolves and Google gives us the tools to do just that.
Join the Discussion