About eight years ago, I came to a frightening realization: To most efficiently assign media to messages, several forms of calculus would have to be applied. It made me shudder, remembering that I’d gotten only three out of a possible five on my AP calculus exam back in high school (and that was probably out of sympathy).
The reality of the media marketplace is that different slots of media are valued differently by different potential advertisers. Thus, setting the optimal price of media involves multivariable, multifunction algorithms that can sometimes be solved only by applying the complicated calculus of integrals.
Worse still, the results of that math will suggest different advertisers should be charged different amounts, depending on their level of desire.
The online advertising industry came up with a dummies’ version of this pricing system by allowing different pricing models. Cost-per-action (CPA) media is sold to people who assign extreme value to highly targeted media. Cost-per-click (CPC) media gets sold to people who highly value targeted media. And cost-per-thousand (CPM) media gets sold to advertisers who assign some value to somewhat-targeted media.
This approach is quite simplistic relative to the potential of customized media pricing. Advertisers wind up giving away a lot of media for prices some advertisers would consider cheap. Likewise, much media goes unsold because other advertisers consider that media too expensive.
Yet another layer of complexity is tossed onto this online media pricing spaghetti: site discretion. When a site does many deals with many advertisers, it faces the question: To whom should it assign the next impression? For instance, should the next impression be assigned to the deal that gives the site $1 per click-through, or should it assign it to the client who gets a lower click-through yet who will pay $2? Or, should the site assign the next impression to a different kind of deal altogether? Of course, the optimal answer depends on how and where the impression occurs, the type of advertising client, and many other variables. It becomes obnoxiously complex.
Google Steps In
That’s where Google stepped in. It just launched a service that allows advertisers to buy CPC media based on their own selection of keywords and a maximum CPC they themselves define. In essence, the advertiser tells Google to give them media with a certain targeting, and Google places them among the sponsored links depending on how much they were willing to pay per click. The system manages to place the ads in such a manner that it closely follows the optimal price for each advertiser, as per the calculus equations.
Anybody with a credit card and a product can set up an account in about 15 minutes on Google’s automated application. You select keywords, write a quick text ad, and define the maximum CPC. Google will give you an estimate of how many clicks you might expect and how much it might cost per day. The campaign can be launched immediately. You are charged only the CPC necessary to land you on top of the listing, provided that cost is below your defined maximum CPC. So, for instance, if an advertiser defines a maximum CPC of $1, and it only takes $0.66 to get on top of the listings, Google will charge $0.66.
My one criticism is the sponsored-links section, off to the right-hand side of Google’s pages, doesn’t get much attention from users. This may be largely parried by the CPC nature of the media deals. If users don’t see the ad, at least there’s no harm done.
The system is so well automated that the application can be used by small businesses that traditionally don’t have enough scale to warrant bothering with online advertising. To test out the system, I set up a campaign to link a series of keywords to my new textbook. I placed a maximum CPC of $1.18. When I tested the system, I was charged an average of $0.71 per click.
The implications are huge. In traditional media, the brunt of spending is done by small companies in media such as newspapers and local radio. We tend not to talk much about that kind of stuff in marketing circles, because we have a bias against small advertisers and because online media haven’t been transactionally efficient enough to bother with them. But, by opening the door to small advertisers, Google may be multiplying available marketing dollars for everyone online.
I called Google’s David Krane to ask a few questions about the service. He emphasized that the system is only one ad channel for the company. “Premium” service advertisers still work with reps.
It will be very interesting to see other sites’ reaction to this approach. Two things have prevented others from adopting a similar strategy. First, sales forces hate CPC and CPA type deals, and they’ve succeeded in making such system development a low priority (if not anathema). Second, the systems necessary to connect trafficking to billing systems to keyword inventory and selection systems contain more software code than it takes to launch the space shuttle. Given the many legacy systems the older sites have collected, creating this type of application is not a trivial undertaking. Google may have benefited in this respect by having been the new kid on the block.
Another interesting aspect will be the implications these systems have on people who buy and sell ads. One could argue that with this type of system (especially if it spans multiple sites), the need for a lot of the work an agency does would become moot. Likewise, the sales force might become superfluous to all but the very largest and vainest accounts.
Unfortunately, many technical decisions made by both sites and agencies often have more to do with the politics of internal departments than with rational factors such as potential sales volume. Here’s hoping the tech guys over at Google (and elsewhere) continue to develop under-the-radar systems. Then, the online media industry won’t stagnate under all those sales VPs in the CPM crowd.