No More Winner Takes All

Over the last year, I’ve been in a number of discussions where the concept of Winner Takes All was raised, and it’s now starting to annoy me. In a Winner Takes All market, there is a dominant competitor who takes a very large share of the profits. An example at the moment is the mobile phone manufacturing market, where it seems Apple is the winner who is taking all (or, at least, most). However, there may be a widespread view that any market relating to the Internet is Winner Takes All, and that would be a problem.

Winner Takes All is typically put forward to justify either betting big (eg. intentionally making multi-year losses in order to get the scale of users/customers needed to be dominant in a market) or not doing anything (eg. because only one can win anything, and the likelihood is that it won’t be you). In other words, Winner Takes All markets are for only the bravest of the brave. But if anything relating to the Internet is Winner Takes All, then unless you’re pretty special, you should stay off the Internet. Or so the thinking goes.

You might expect that I disagree – and you’d be right. Let me break down why.

Firstly, there are mature businesses on the Internet that have multiple big players, and yet not a single winner. Web mail is a good example, with the biggest services (from Microsoft, Yahoo and Google) having similar sized user-bases.

Global Web Mail Unique Visits (ComScore May 2012)
Service Users
Microsoft Hotmail 325 million
Yahoo Mail 298 million
Google GMail 289 million

And while one counter-example is enough to disprove a hypothesis, here’s another one to show the first wasn’t merely an exception. Desktop web browser share is largely split between three big players (Microsoft, Mozilla and Google). Another one is Internet-connected game consoles. I’d say this myth is busted.

A response to the above is granting that not every market relating to the Internet is applicable to Winner Takes All, but that there are some important ones that are. For example, Internet services with “network effects” (those where the more users that adopt it, the more value they are to those users) are in such a market, and Facebook’s dominance in social networking illustrates this.

While this watered-down Winner Takes All view appears more reasonable, there are two lines of evidence that discount it also. The first is the historical record of all the previous social networking services where it appeared there was a winner, but then they lost to a subsequent service that rapidly took over. Back in 2007, MySpace was considered dominant over Facebook, and before MySpace were other services like GeoCities which, according to Wikipedia, in 1999 was the third most visited site on the Internet. If a winner can be displaced so quickly, can they really be said to have “won”?

The second line of evidence is the active competition still occurring in the social networking market. There are both alternative services such as Twitter, LinkedIn and Yammer, and also similar services operating in specific (yet still sizeable) markets such as Qzone, Renren or Sina Weibo in the Chinese language market. If a service isn’t dominant everywhere, can they really be said to have taken it “all”?

But wait, I hear someone say. Cyworld was dominant in the South Korea language market, and yet now Facebook has displaced Cyworld over the course of a year. Doesn’t this show that the same could happen in China and Facebook is operating in a Winner Takes All market? Well, yes it could happen in China, but no, all this shows is that Facebook is a good competitor. There’s no need to explain away Facebook’s appeal through claiming their rise in South Korea was an inevitable consequence of the market structure.

So, I don’t find Winner Takes All convincing, but the danger is that some people believe it and choose not to attempt to launch valuable Internet-based ideas. We users of the Internet would end up deprived of those services as a result. But, it seems the good news is that plenty of people do not believe in the pessimistic world view of Winner Takes All and are happily putting their products and services on the Internet.

Book Review – Good to Great

I’d been aware of this book for a while, but it still seems to be available only in expensive hardback format, so I was waiting until it got cheaper. Recently I found it for $15 (still hardback) and this was enough for me to give it a go.

Good to Great

Research-based guidance for established companies to excel in their markets.

I came to this book by Jim Collins with some interest in reading about a new research-based attempt to find a winning corporate formula, but also scepticism due to the unsuccessful attempts that have come before. Perhaps the most infamous was In Search of Excellence which purported to find the recipe for excellence, but gave Atari (had to sell key assets in 1984) and Wang Labs (filed for bankruptcy in 1992) as examples of excellent corporations. Although, that book identified 43 “excellent” companies, so it’s probably not too bad for only a couple of bad apples to end up in their list.

Collins improves his odds by identifying only 11 “good to great” companies. But this is perhaps an uncharitable comparison, as his team appears to have done an extensive job in analysing these companies, and there are only 11 because only 11 companies out of the 1,435 US-based “Fortune 500” companies from 1965-1995 met their criteria. Then to identify the features that relate to being “good to great”, these had to be possessed by all “good to great” companies and lacked by all 17 close-but-not-quite-good-to-great companies also identified by the team.

The book explains the basis for these features, and is engaging and well-written. For me, the most surprising was the feature of “first who.. then what” which is basically the idea that hiring well becomes foundation for all corporate strategy, and not, say an analysis of competitors, technology, financials, or other market fundamentals. I do like this idea, despite its fuzziness, as it says that people aren’t fungible and that they can make a big difference. There are five other features, making six in all, but none were as counter-intuitive as this one. In any case, I will now be paying attention to these features in my workplace and future employers.

However, I can’t bring myself to adopt them as fundamental tenets since despite the rigorous research, the conclusions remain essentially unproven. From my point of view, there are three weaknesses in the research: the set of “good to great” companies is arbitrary, the set is small, and the conclusions are untested.

Taking the first problem, “good to great” companies were defined as having a transition to “great” performance of at least three times the general market (from a point of transition). If, instead of three times, it had been five times or even two times, a different set of companies would’ve been found. Since the features needed to be possessed by all “good to great” companies, a different set would’ve produced a different set of features, e.g. potentially larger or smaller. Hence, perhaps the features found are sufficient for a good-to-great transition but some weren’t actually necessary.

The problem of a small sample is tackled in the book, referencing “two leading professors” who think the sample of 11 companies wasn’t small. Unfortunately, this is not convincing. For example, one professor says that the 11 companies wasn’t a sample as it was 100% of companies that met the criteria – although I would respond that the book promises that these principles are universal, so there will be more such companies in the US-market in the future, and they should also apply to non-US-based companies, hence the 11 companies don’t represent 100% of all possible “good to great” companies.

Lastly, the conclusions are untested. The research team could’ve, say, looked for a couple of companies outside the US that met their “good to great” criteria and then checked that those companies possessed all of the six features. Except they didn’t. The only companies examined as part of the study were those that informed the conclusion. The use of comparison companies gives me a level of faith in the conclusions, but these can’t be validly re-used in testing that conclusion. So, really the conclusion remains a hypothesis for now.

My grumblings notwithstanding, I was impressed with the analysis in the book and the methodology that used comparison companies to filter out features that were shared by both the “good to great” companies and also those that didn’t perform so well. It has shifted my thinking about what a successful business can look like.

Rating by andrew: 3.5 stars