This is the first of four-posts in which I want to explore the possibility of creating and bringing to market a disruptive search technology: how strong are the natural barriers to entering the market, what “inning” of search technology are we in, how effective is google’s search technology anyway, and what could a disruption in the industry look like?
Benchmark Capital VC Bill Gurley recently wrote a well-followed article discussing Google that was picked up by Tech Crunch under the apt title “Search is a castle, everything else is a moat.” Gurley’s well-taken argument is that Google subsidizes peripheral products like web browers, gps navigation systems, and mobile phone operating systems in order to protect its money-making search business by controlling potential distribution channels. Gurley’s argument is certainly relevant to anyone trying to compete with one of google’s subsidized arms. In this post, I will attempt to analyze how relevant it is to competitors in google’s core search business. Is this “castle” really as unassailable as Gurley and other prominent investors seem to think?
As a new economy titan, google is often assumed to have some kind of special web enabled “network effect” that prevents others from competing with it. Too often, the exact nature of this supposed barrier is left un-described. On close examination, it is hard to find evidence of any strong network effects in the google model.
- Direct network effects occur when the value that accrues to each of a company’s customers is proportional to the number of customers. Strong direct network effects can result in a winner-take-all market since smaller players are unable to provide much value. Social networks like Facebook are an obvious example – the network has no value unless a critical mass of your friends are on it. In contrast, the value a search engine provides is entirely dependent on the quality of the results, with no direct bearing on the number of other users. If blekko can get you to the data you want faster than google, why would you use google just because everyone else is?
- High switching costs occur prominently in businesses that are platform-based or have a high learning curve. Complex enterprise software systems like Oracle are one example – to switch to a rival system often requires re-training an entire organization, at significant cost. In contrast, it is completely trivial to switch between doing searches using Google and Bing – just type a different address or change the default search option for your browser.
- Multi-homing costs occur when the cost of running two systems in parallel are high – for instance, using both Apple and Microsoft may result in purchasing redundant software. Yet again, it is entirely trivial to run multiple search engines – you can even run them at exactly the same time in side-by-side tabs and compare results. This makes it easy for users to trial a new search engine with no cost before they decide to switch their default choice.
- Indirect network effects occur in platform business when the value to one side of the platform is dependent on the number of users on the other side. The rise of Windows owes to this phenomenon – users purchased the OS because the majority of software applications they wanted were developed for it; developers wrote applications for Windows because the majority of users they wanted to reach used it. In contrast, google’s indirect network effects only go one way. Advertisers are drawn to google as a result of its strong user base, but users do not care much how many advertisers are on the system.
In contrast to its new-economy brethren like Microsoft and Facebook, Google’s real source of competitive advantages feel very industrial-age – scale and brand.
- Much like building a new auto plant, the costs of entering the search business are quite high since a) The technical challenges of building a modern search engine mean you need a significantly large amount of capital to hire the necessary engineers/scientists and build the required data centers and b) ingrained consumer habit/inertia (e.g. to “Google” has become a verb) significantly lengthens the likely market penetration rate and increases the chance of an expensive failure.
- The indirect network effects discussed above amplify scale and brand effects because Google is able to subsidize users by charging advertisers, while any new market entrants would be unlikely to attract any advertisers before building a critical mass of users and would thus be forced to either: a) Charge users to search or b) raise additional capital to absorb initial losses.
- A variety of other arguments for google’s strong assumed position are bunk.
- Some argue that the quality of searches improves with the number of searches done since google can presumably data-mine its results. But this is true of any business – there will always be some learning curve and return to experience. It is a scale effect – and a weak one at that – not a network effect.
- Some argue that consumers benefit from the presence of advertisements, making the indirect network effects go both ways. If this was true, we should expect to see much higher click-through rates on ads. It is marginal at best.
- Some argue that google is dominant because it hires the top talent engineering talent. But this is an operational advantage that can change fast – witness Faceook’s ability to poach top talent.
A few conclusions come from this:
1) While still strong, the natural barriers to Google’s search business are less than popularly perceived. If this is a castle, it is a castle on a hill – not on a mountain.
2) Successful “moat-building” by google will at most cut off one of two routes of attack on its castle. Starting a next generation search engine will require either a) Raising a large enough sum of money from VCs to develop search technologies and withstand early losses while attracting advertisers and building a brand b) using an existing peripheral business as a source of both cash-flow and distribution for an emerging search engine. By knocking off other competitors, google reduces some routes of b.
3) However, the effect of “moat-building” is necessarily limited since:
- They can never cut off all potential routes (witness facebook, IOS, Quora, Yelp, etc.) and
- Even if they could, so long as there is a free internet and users can go to whatever search site they want, google can never kill off the chance of a brand-new entrant with a disruptively better technology.
In summary, for all the talk about constructing moats, the primary barriers holding up google’s search castle basically come down to the age-old phenomenon of scale and brand. Not so different than Wal-Mart or General Motors.
At least that’s how I see it.