Relevant Market for Assessing Google Power
One of the first questions addressed in any antitrust inquiry is how to assess the relevant market. You need to know what market a company is in before you know if it dominates it, or if it enjoys monopoly power. This is not always obvious. Does Kraft dominate the market for packaged macaroni and cheese? Or is the relevant market inexpensive and easy-to-prepare shelf-stable kiddie-lunches, in which case Kraft shares the market with Campbell's Tomato Soup and Bumble Bee Tuna and Skippy Creamy Peanut Butter.
When Google defines its business and estimates its market share, it declares itself to represent approximately 3% of all advertising sales in the US, or alternatively as about 30% of all online advertising sales. It does acknowledge a share of paid search in the US of approximately 70%, but argues that this is not an interesting or significant statistic, since search is not a business segment and advertising and online advertising are businesses. A 70% market share is within the range that that the Department of Justice uses to assess a rebuttable presumption of possession of monopoly market power. It is well above the level of 27% at which Apollo (United Airline's Computerized Reservation System) demonstrably possessed monopoly market power in the 1980s. Importantly, at that time Apollo employed a business model similar to the business model Google uses today.
So it matters which industry definition and which market share figure we use. Is the relevant measure of Google's market share 3% of advertising, 30% of online advertising, or 70% of search? This determination will dictate much of the subsequent course of any antitrust litigation that Google may face after the FTC investigation.
The question of relevant market is almost always essential to address and to address correctly in an antitrust dispute. If AT&T is allowed to acquire T-Mobile they would represent a vanishingly small fraction of the planet's technology sales, a larger but still non-problematic share of the planet's total sales of cellular bandwidth, and a significantly larger share of America's total sales of cellular bandwidth. (Interestingly, the Department of Justice has filed an antitrust lawsuit to block the acquisition, although the resulting company would still have a significantly smaller share of wireless communications than the share of search enjoyed by Google.) At the time of Microsoft's antitrust litigation, the company was likewise a vanishingly small share of the planet's technology sales, perhaps 3% of America's share of software sales, and perhaps as high as 90% of the sale of operating systems for Intel-based PCs. Which number should the DOJ have used when assessing Microsoft's market power, 3% or 90%? And which number should the DOJ use when assessing Google's market power today?
How would you know what market to use?
How would you know what the relevant market is for any product?
A relevant market is defined according to both product and geographic factors. In general terms, a relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable (substitutability) by reason of product characteristics, prices and intended use.1
In determining whether a hypothetical monopolist would be in a position to exercise market power, it is necessary to evaluate the likely demand responses of consumers to a price increase. A price increase could be made unprofitable by consumers either switching to other products or switching to the same product produced by firms at other locations. The nature and magnitude of these two types of demand responses respectively determine the scope of the product market and the geographic market.2
In the case of the AT&T / T-Mobile cellphone acquisition, you can use a Verizon cellphone plan as a substitute for an AT&T plan, but you cannot use a Comcast landline as a substitute for mobile telephony. Likewise, you cannot use an Airtel wireless plan from India or a Vodaphone wireless plan from the UK as a substitute for a domestic American wireless plan (at least not cost-effectively). The relevant market for assessing the AT&T / T-Mobile deal is therefore other domestic US mobile cellphone service providers.
When assessing Microsoft's market power in operating systems, you could have used Linux or Red Hat as operating systems for your Dell or Lenovo machine, but you (obviously) could not have used mainframe operating systems for IBM big iron, or Adobe Photoshop or other non-OS software, or even Apple's OS X. The relevant market was limited to other operating systems for Intel-based PCs, and Microsoft's share was indeed close to 90% of this relevant market. How would you determine the relevant market for Google? Is search a form of advertising or is it a form of distribution? Does Google sell ad placement? Or does Google sell access to customers?
How are these different? An ad is generally defined somewhat like the following:
The non-personal communication of information usually paid for & usually persuasive in nature, about products (goods & services) or ideas by identified sponsor through various media. (Arenes 1996)
That is, an ad is a message, perhaps a few lines or even a few pages in a newspaper or a magazine, or a few seconds or a minute on radio or television, that (1) seeks to change my opinion, and thereby (2) change my behavior, by (3) changing what I want to do, or what I want to buy, or how I want to vote.
In contrast, paid search is a form of selling customer access. After seeing an ad or watching a TV commercial I may decide to book a Marriott, but if I am redirected through search I may decide to reserve an Intercontinental or a Westin instead. Search is not a bad business model, but it is not the same business model as advertising. Search can do more than substitute for an ad, it can trump an ad.
How are ads different from paid search?
Let's see why search is not just another form of advertising, using a sequence of progressively more subtle examples.
Hotels.com is clearly a distribution system, and not an advertising system. When a hotel refused to give hotels.com a sufficient supply of discounted room, hotels.com would list the hotel not as unavailable but as sold out. Individuals who searched for a property on hotels.com would either book or conclude that it was sold out and book someplace else. The power to redirect customers clearly is separate from and in some cases is more powerful than the brand image created through advertising.
The airlines' computerized reservation systems (CRSs), which were provided to travel agencies, were likewise distribution systems and not advertising systems. When Sabre relocated flights from Braniff deep in the search list, not on the first screen, and when Apollo did the same to flights from Frontier, both airlines ended up having to file for bankruptcy. A passenger might have requested a Braniff flight from Dallas Fort Worth to Chicago O'hare, but if the passenger's travel agency could not easily find Braniff flights, the passenger was booked on American or United. Once again, the power to redirect customers clearly is separate from and in some cases is more powerful than the brand image created through advertising. Indeed, the nature and purpose of these systems is clear, and they are now called Global Distribution Systems (GDSs) rather than CRSs. They were never called advertising search systems.
So what about Google? Is search a form of brand building and advertising, or a form distribution system like hotels.com, Sabre, and Apollo? Clearly a bit of both. A client with a well-known hotel brand and a huge budget for traditional advertising once complained that unless he participated in keyword auctions he vanished from the top of the page even when travelers searched specifically for his hotels. He reported that every dollar he saved by refusing to buy his own keyword back cost him between $30 and $40 in lost revenue. Similar claims are equally hard to document, but have been made by hotel operators, airlines, and retailers. The power of redirection does appear to trump the power to advertising.
Why do we care?
With a market share of 3% in advertising, we really should not care about potential monopoly power at Google. With a market share of 70% in search we should care very much indeed. Why do we care what market to use when assessing Google's power? Because power can be abused, and power over search is enormous power indeed.
Well, yes. Google now calls its top spots Ads instead of Sponsored Links and continues to earn virtually all of its revenue from Adwords and Adsense. It's hard to be more clear than that. Abe Lincoln is said to have asked "if you call its tail a leg, how many legs does a cow have?" Not surprisingly, the correct answer remains four. Calling a tail a leg does not make it a leg, and calling something an ad does not necessarily make it an ad.
But how would you know whether Google was an ad company or something very different? What data should the FTC attempt to get? What should it consider when determining whether Google is in the advertising business or the business of selling access to customers?
This determination will be critical in assessing Google's power, in determining if antitrust litigation is appropriate, and in structuring antitrust litigation should it occur. The remaining questions about having or abusing monopoly power will be assessed separately.