Of Search Engines, Ad Prices, and Blind Sales

A report that Google and Yahoo are losing ad sales to an upstart might, at first glance, raise doubts about the search giants’ understanding of the advertising business. However, a closer look suggests the famously smart people at Google and Yahoo know exactly what they are doing.
The article in Monday’s New York Times reports a new company, Quigo Technologies, convinced Fox, ESPN, and Cox Enterprises to abandon the search giants as providers of text advertising. These ads appear beside your search results in response to the words you are searching for.
Quigo, unlike Yahoo and Google, tells advertisers where ads appear on the web and allows advertisers to buy ads on specific sites, the Times reports.
It may seem odd for search giants to conceal this information because advertisers target people likely to buy their products. So most advertisers want to appear on websites with content that attracts large numbers of people potentially interested in their products – if you sell sporting goods, you want to be on the big sports sites.

How Search Advertising Works

According to Advertising Age, the search giants make most of their money from these text ads.
Advertisers buy key words, and their ad appears when those words are typed in a search engine. The advertiser only pays if someone clicks on an ad. Advertisers bid a certain amount for each click, and that determines how often and where their ads appear.
The search giants display the ads on their pages and on innumerable other web sites participating in their advertising programs. The advertiser knows how many clicks it paid for, but not where the clicks came from.
If advertisers knew ads were appearing on sites they want to target, wouldn't they buy even more ads? So why is this information concealed?

Concealing Information Increases Ad Revenues

I am indebted to Roy W. Kenney and Benjamin Klein , authors of a 1983 article in the Journal of Law and Economics (cite below), for this explanation:

Advertisers consider some web sites to be better than others. They may be attracted by the large number of sites available through Google and Yahoo, but most advertisers would probably like to select some sites and ignore the rest.
But advertisers cannot select sites, so they must instead estimate the average value of a click from all sites, desired or not. Many people who click on an ad will not buy anything. And sites with small amounts of traffic probably have smaller proportions of buyers among those who do click on ads.
So the best sites will generate large numbers of clicks and many more likely buyers. Another way to think of this is that average sales per click will be much higher on the desirable sites.
If advertisers could identify the desirable sites they would bid more for those, and advertising would be concentrated there. But Google and Yahoo would be forced to sell ads for much less on the undesirable sites, and their total advertising revenue would decline.

Two Alternatives

What if Google and Yahoo instead offered to sell advertisers web sites in groups -- or bundles -- and included some less desirable sites in each bundle? Buyers would look for bundles they considered bargains, and only bid for those. The search giants would be forced to sell ads on the remaining bundles for less than buyers would pay if they did not know what bundle they were getting.
Another alternative would allow buyers to experiment, buying ads on a variety of sites to see which sites produced the best results. Advertisers would make adjustments after their initial purchase, asking the search giants to redirect ads to sites generating the most sales. But this would be costly for the search giants. It would also have the same effects on ad prices and revenues as the other alternatives.

A Change is (Probably) Gonna Come

The existing arrangements also make it possible for Google and Yahoo to distribute ads differently than expected. For example, an ad on 1,000 blogs might generate 1,000 clicks, but few buyers. The same ad on a few big news sites might generate the same 1,000 clicks and deliver far more potential buyers. Google and Yahoo are asking buyers to trust that this is not happening.
The Times article also reports some buyers are worried about click fraud. This happens when people click on ads just to generate revenue for the sites where they appear (Google and Yahoo share some of the revenue with the site).
The article says Google is planning to give advertisers more information about where ads appear. The search giant is feeling competitive heat from companies like Quigo. It will be interesting to see what happens next.

(The cite for the Kenney and Klein article, "The Economics of Block Booking," is Journal of Law and Economics, Vol. 26, No. 3. (Oct., 1983), pp. 497-540.)

Competition and the satellite radio merger

Prof. Stephen Lacy of Michigan State University is this blog's first guest commentator. He has this response to my Feb. 21 suggestion that prices will be higher if the two satellite radio companies are allowed to merge:

"The particular blog prompted an idea about substitutes. Satellite really has two submarkets -- in the home and in the car.
In the home, Sirius and XM compete with the three-dozen or so music channels on digital cable, iPods and with specialized Web sites such as Pandora.com. In the car, the companies compete with iPods, books on tape and CDs.
These are all pretty good substitutes for music, which suggests satellite will differentiate itself with programming other than music. In other words, the potential price increase might be acceptable because of programming that might not otherwise be available."

(Full disclosure: Prof. Lacy was my dissertation advisor.)

Things to consider about satellite radio

The proposed merger of the satellite radio companies, XM and Sirius, is getting a lot of attention because of the size of the deal (Bloomberg reports its worth $4.84 billion), because of uncertainty about how federal regulators will view the deal, and because consumers might be hurt by the reduction in competition.
The companies have about 14 million subscribers between them, each paying about $13 a month or more. A rough estimate suggests the merger would result in a company with subscriber revenues of at least $2.18 billion a year. However, the Chicago Tribune says the companies lost a combined $7 billion getting started, and the new company would have about $1.6 billion in debt. The companies argue the merger is critical to their survival because of intense competition from programming on the Internet, cell phones, and broadcast radio.

Some questions for regulators

Regulators at the Federal Communications Commission must approve the deal, or it won't happen. It won't be easy for the FCC to evaluate the arguments that consumers are substituting all of those alternatives for satellite radio.
First, consumers pay to receive satellite radio, but not for advertiser supported radio or Internet programming. But the best way to tell if goods are substitutes is by measuring the relationship between the price of one good (satellite radio) and the demand for other goods (everything else). How do you measure the influence of price if consumers can get the substitute goods for free?
Second, the market for earthbound broadcasters is defined by the reach of their radio signal, but the market reached by a satellite is not limited this way. And there is no good way to define the geographic extent of a market on the Internet.
I'm skeptical that all of the other media being cited by the companies are substitutes -- many people who consume free radio would not be willing to pay for the same programming. This does not mean the companies' arguments lack merit, but it's in their interest to push their definition of the market to the limit.

What about consumers?

There are two possible ways consumers could be hurt. First, the merger probably will result in higher prices. Second, the range of programming might be reduced.
The second point first. Phil Rosenthal at the Chicago Tribune points out that XM and Sirius offer different programming -- one has Howard Stern, and the other Oprah, one has football, and the other has baseball. So customers who like particular genres, such as sports, have to subscribe to both XM and Sirius to get all of the programming they want. However, it's unlikely many people are actually willing to pay for both services, not to mention impractical.
Customers who want programs on the service they don't subscribe to would be better off if the merger goes through. The new company will probably offer all the programs now offered separately because elminating programming means the new company would lose customers.
As a general rule, increasing the range of programming at a merged company would increase the number of subscribers. For instance, sports fans and opera fans would subscribe if they could both find programs they liked for a price they were willing to pay.
This goes back to the first point. Prices will probably increase, but subscribers will probably get more programming. Customers should pay more if they are getting more.
But will the new prices be so high that they cannot be justified by the increased programming?
This might be the outcome if the new company can monopolize its segment of the radio market. And that is why the market definition that the FCC adopts will be so important to this proposed deal.