Google's two sided markets

I was thinking about Google last week. One of my student teams did their report on Google (a popular topic around here). And the author of a Google book wanted to interview me about platform issues.

The students did an analysis of the search industry. Oversimplifying, they concluded that Google would have trouble making money because Google’s customers have high bargaining power due to low switching costs — because they can switch to another search engine at any time.

I had three problems with this. The obvious one is that Google is making money: last year, $4.2b net on income of $16.6b, or 25.3% net after. There are possible explanations for this contradiction, the first being that you can’t use the industry’s market share leader (and most profitable firm) as a proxy for the entire industry. We don't know what MS or AOL are making in search, because they’re too diversified, and Yahoo’s profits has shown wild swings in the past 4 quarters, from 3% to 30% net after.

Still, I always tell my students to check their industry analysis against industry profitability for consistency, so this big discrepancy was unsatisfying without being a conclusive proof of something wrong.

The second problem was that people could switch, but they don’t. Why not? Habituation — psychic switching costs — is an explanation, one consistent with the psychic costs I saw in my dissertation. But the switching costs between Google and Yahoo are 100x lower than between Microsoft and the Mac, and yet Apple is gaining share and Yahoo is not.

Which brings me to my third issue — the only one that produced a satisfying answer. Suppose users could switch — and they did? Would it affect Google’s profitablity? Of course not.

Google gets its revenues from advertisers, not users. It has a two-sided market (or, perhaps more accurate, a two-sided platform) supplying content to users and user eyeballs (or clicks) to advertisers.

So as long as Google does a very good job of delivering the right users to advertisers, the advertisers will have high switching costs and will stick with Google. So Google will continue to monetize its users better than its rivals.

Bad week -- and decade -- for newspapers

It was a week of bad news for the U.S. newspaper industry, just the latest installment of a decade of such bad news.

• On Monday, a study by former LA Times reporter Tyler Marshall documented the downward spiral of the nation’s 1,217 daily newspapers. After interviewing executives from 15 papers and surveying 259 newspapers, the study from the Pew’s Project for Excellence in Journalism showed that executives are cutting staff, national and international news, but don’t have a real plan to turn things around.
• On Wednesday, the industry’s premier property (The New York Times Company) announced dismal earnings. In just a year, the company’s stock price has fallen to half its earlier value, to the point that the flagship paper is valued nearly the same as its suburban rival, the much smaller Long Island Newsday.
• Finally, on Thursday, the San Diego Union-Tribune, the oldest business in my hometown of San Diego (and the second oldest newspaper in Southern California) went up for sale. Apparently his financial advisors convinced owner David Copley that he could more easily support his lifestyle if his wealth was in T-bills rather than an illiquid, privately held newspaper descended from papers his grandfather bought 80 years ago.

The problem for Copley — like other sellers — is that there are few buyers. Once upon a time, Copley might have hoped for a merger with one of the two nearest newspapers: the Orange County Register or LA Times (which ran a SD editon from 1978-1992). However, both are in trouble: like most family-owned newspapers, the Register has unhappy heirs that care more about money than journalism; the LAT went through this already, resulting in the paper being bought in 2000 by the Tribune Company (and being badly run ever since). Among other likely buyers, national newspaper chains like the Times Company, Gannett, and Media General are reporting declining earnings while struggling with high debt from previous acquisitions.

Of course, stress and realignment for the industry come with every major dip in the economy, when fewer housing sales and reducing hiring squelch the two major sources of advertising revenues. In California, the 1992 recession killed the San Diego Tribune and the LAT’s SD edition. I bailed out of the newspaper business at the end of the 1982 recession, only six months before my coworkers traded up to the bigger paper (and a $100/week raise) when classified ads rebounded.

However, in 2008 the real pressure is coming from long-ignored structural problems. Newspapers are the ultimate information good — lots of up front costs and near-zero marginal costs (even less than for software, which tends to generate support costs proportionate to the quantity sold).

So, as with any other network effect, the virtuous cycle on the way up (success increase profits which allows a better product which fuels success) becomes a vicious cycle on the way down (declining sales means cuts reducing product quality which lowers sales further). Of course, I watched this vicious cycle nearly put Apple out of business, first as an Apple ISV and then in my dissertation.

Newspapers have a particularly hard problem — escaping the complacency of the past four decades. With few exceptions, all the major newspapers in the US were either a monopoly or cozy oligopoly (usually a duopoly). They didn’t have to compete to make money, they just told advertisers “if you want to reach local buyers, we’re the only game in town” and (at least before television) told readers “if you want local news, we’re the only game in town.” (In two-sided market terms, this is called charging both sides, although advertising provided 70-80% of the revenue.)

Now, of course, we have at least a decade worth of young people that no more expect to pay for news than they would pay for music; somehow they still pay for movies, but that probably won’t last either. The papers have reached out to new readers with their online websites, but as U-T investigative reporter David Hasemyer dryly put it,

While many newspapers' Web sites have recorded jumps in readership, media analysts have pointed out that advertisers generally don't consider an online reader to be as valuable as a print reader.

Even for those newspapers that enjoy broader reach than ever — notably the New York Times — a shift from print to online means that revenue (and profit) per reader has plummeted.

Newspapers face two structural problems and have been unable to fix either one. Due to new web-enabled competition, newspapers have lost their pricing power both sides of the two-sided market — and there seems no turning back.

First, is there are so many other ways to reach readers via the web — whether CraigsList, Monster.com or Google’s localized version of AdSense. The cost of one impression has plummeted from the days of dead trees when there was only one regular way to reach a given household.

More seriously, online information is now a commodity: people get news free now from so many online sources. (One major sources is from Google and Yahoo news sites, that use AP news compiled from local papers). OK, the NY Times or the big city daily has better news, but how much better? If it’s $20/month (or even $10 or merely requires a login) will readers bother? Most won’t. As with other commodities, better loses to “good enough.”

Some newspapers think it’s just a matter of getting a good revenue model, or fixing website design to be more compelling, or some other tweak. After all, they argue, people want local news and there’s no other source. This argument has been made by insiders for years.

To this, I’d say: “where’s the existence proof?” The newspaper industry is notorious for copying each others’ design, circulation and advertising tips, so with 100+ newspapers with more than 100,00 circulation, one of them should have figured it out by now. Instead, newspapers are using the same online revenue models they always have.

The only exceptions are financial newspapers like the WSJ and FT (where business owners and investors are willing to pay a premium for slightly better information) and the free suburban weeklies and subway dailies (which pay little or nothing for content, and it shows.)

Of course, most journalism students (like poli sci or history students) know nothing of economics and little of business. (BTW, back in 1981-1983, that included me too). So they keep hoping that — since all their friends know they have a better product than the bloggers and the wires and the other free sites — someday the tooth fairy will come waive her wand and everything will be right. But it ain’t gonna happen.

For nearly a year, rumors have been swirling that a big city newspaper (such as the SF Chronicle) will stop killing trees and go online-only. The first one will get hit with a double whammy: someone else will swoop in and get the dead tree advertisers (in this case, the SF Examiner or the Bay Guardian or the Oakland Tribune) while the collapse of revenues will mean a cutback in the news staff and thus the quality of the online product. Maybe the paper could stabilize with an editorial staff of 150 instead of 500, but (as with any newspaper today) the process of downsizing would be bloody, shaking the confidence of employees, and both types of customers.

Some rich guys buy money-losing sports teams for ego reasons; perhaps someone will do that with a newspaper. The problem with this sugar daddy fantasy is that the pro teams are an appreciating asset, while newspapers are clearly a declining one. Also, there are no luxury boxes in the newsroom that billionaires can use to impress their friends.

Platform strategies: the academic view

Tuesday night I got back from a quick trip to London, which meant a lot of time in planes and five nights arguing with my body (mostly losing) about when (and when not) to go to sleep.

The reason I got on the plane (other than I could) was for a conference at Imperial College called “Platforms, Markets and Innovation.” The conference was organized by Annabelle Gawer, a lecturer at the Tanaka Business School. For us yanks, TBS dean David Begg noted Imperial’s role as England’s leading school of technology: 50% medical school (largest in Europe), 25% engineering, 20% science and 4% business.

Annabelle has spent the last decade studying the business of platform management, which was the topic of her 2000 dissertation at MIT which became a 2002 book from HBS Publishing.

I’ll admit I was a little slow to warm to Annabelle’s conception of the platform. But after reading her excellent 2007 article on Intel’s platform strategies (with Rebecca Henderson), I blame it on HBSP. Their formula for selling managerial books encourages (nay, requires) overclaiming and broad assertions that the book solves all the world’s problems — past, present and future. (For example, Clay Christensen’s excellent 1997 book suffers from this problem.) Meanwhile, academic papers (through the peer review process) tend to be more modest about claims that can be directly supported by the evidence, at least until you get to the part marked “implications.”

Annabelle and I are among the few people who’ve focused our academic energies on IT industry platform competition. We both build upon the masterful 1999 paper by my friend Shane Greenstein (and his former dissertation advisor Tim Bresnahan) which explains the success of computer platform strategies from 1965 to 1995. My own 2000 paper (with Jason Dedrick) on PC platforms emphasizes the role of technical control of a platform, while my most cited work (a 2003 paper on open source) is one of the first first academic papers on open platform strategies. (A 1993 paper on Sun Microsystems by Raghu Garud is about open platforms but doesn’t call it that). But in comparing myself to Annabelle, she is has clearly devoted her full energies to platforms over the past 8 years, while I’ve also been doing research on open source and open innovation.

Monday’s conference brought together the best platform strategy researchers in the world, including a lot of jet-lagged Americans as well as an equal number of European researchers. The goal was to air and discuss chapters from Annabelle’s forthcoming edited volume called (surprise!) Platforms, Markets and Innovation.

From a 15 minute PPT deck, it’s not always possible to understand the heart of an intellectual argument. Also, I can’t really capture in a 750-word blog posting 12 papers presented over a six hour period.

Instead, let me highlight those papers that offered new insights for my own work on platform strategies:

• Annabelle’s introduction to the conference and the book presents her latest refinement of what platforms are and are not. (Unlike some theorists, she is going to great lengths to drawn boundaries.) Her current interest is in studying the link between platform design (and strategy) to industrial platform dynamics.
• Jason Woodard of Singapore Management U. (with his advisor Carliss Baldwin of HBS), who presented very intriguing work to define a more precise way to represent platform architectural relationships — and to use that to predict which parts of the platform will have competition and which parts won’t.
• A mathematical model by Geoff Parker and Marshall Van Alstyne about third parties adding value on top of the platform — and when should that layer be subsumed into the operating system to provide a building block to others. (E.g., OS X bundling a database, Windows bundling a browser).

Of great personal interest to me was the paper by Shane Greenstein on openness in the Internet. This is picking up a thread of a conversation between us going back to 2004, when he invited me to present a paper at his conference on standards policy. That conference paper became a chapter in his 2006 book with Vic Stango — probably my favorite book chapter of the 11 I’ve authored thus far. (The 2006 chapter on open standards and the 2003 paper on open source strategies are the two pillars of my research on open IT strategies).

In his presentation Monday, Shane contrasted the two great incipient platforms of 1995 and asked whether openness mattered. His conclusion was that openness made no difference to large business vendors (the IBM and MCIs of the world) who would have participated in a closed Internet, but it did keep IETF volunteers active and energized. On the negative side, since the platform isn’t owned, there’s no obvious leader to orchestrate a rollout (as with the August 1995 “start me up” celebration). I am hoping that in the final version will bring out the contrast between IETF and W3C, since both fit most definitions of openness but corporations were much more prominent in W3C.

As I said, this brief summary is not enough to capture the intellectual heft of the arguments: buy the book! As someone who did a 2000 dissertation on standards competition, I thought the field was going to fade away. But Greenstein is continuing the work in economics, while Gawer, Baldwin and Woodard (among others) are developing new insights with platforms. Meanwhile, even “two-sided markets” are telling us things we didn’t know before about standards and platforms.

The session concluded with an industry-academic panel discussion of platform strategies. More on this later.

New two-sided wineskins

For the past few years, I’ve been hearing a lot about “two-sided markets”. There’s been a series of well-cited papers by Jean Tirole, there’s now a book and even some of my friends (like Tom Eisenmann) are getting into the act.

To me, two-sided markets looked like “old wine in new wineskins”: from what I saw, calling it a “two-sided market” doesn’t change the phenomenon or what we know about the role of complements in providing value. We already knew about the hardware-software paradigm, network effects, ecosystem management, and a range of other issues about managing the creation of value for IT systems. Much of this was published by the greek-letter economists (name like Katz, Shapiro, Farrell, Saloner) in the top economics journals (and at least one book).

The two-sided literature emphasizes getting money from one party while encouraging (or even subsidizing) another party. However, cross subsidies are also old hat. For more than two decades platform vendors (like Apple) have been subsidizing their developer relations program from sales of their products, while for nearly that long videoconsole makers have been selling consoles at cost while exacting a tax on videogames.

Sunday night, in a London hotel before a conference, I ran into one of the leading economists studying two-sided markets, Geoffrey Parker. Geoff and I met back in February 2000 when I visited Tulane for a job interview. Among other things, Geoff is a co-author with Eisenmann of several papers on two-sided markets.

So at the Regency Hotel, I cornered Geoff and pointedly asked: “what’s really new about two-sided markets?” Geoff was equally honest, and said there are three main points

1. The two- (or multi-sided) market emphasizes that you can charge for any part of the value created, as long as you charge somehow. If they are “sides” rather than “product” and “complements” you don’t make an a priori assumption about which part is the paid part.
2. As with any good economists’ breakthrough, it provides a mathematical way to find the optimal strategy — in this case, the joint optimization of the returns from both (or all) sides of the market.
3. Antitrust regulators dislike predatory pricing (selling products below cost), but total consumer welfare may actually be better off with cross-subsidies (or at least no worse) if it raises volumes and thus economies of scale.

In other words, to optimize your returns from a two-sided market, you charge more either for the bigger market or where the demand is less elastic.

On these modest claims, I’m now sold. Two-sided markets provide an incremental improvement over prior knowledge, and I will cite it in my own work. Otherwise, it’s just econometricians and game theorists crowing about having greek letters where none went before.