Historically the defensibility of information technology companies has been based on intellectual property – usually patents, sometimes copyrights. From the late 70s until the late 90s many venture backed start ups contributed key components to the information technology infrastructure we all now depend on. These businesses required a big R& D investment and often worked for several years developing their product before they hired their first sales or marketing people. Once they were ready to go to market they distributed their products through major equipment companies that sold a broad range of products to a large base of customers. Because it would have been relatively easy for a major manufacturer to mimic the innovation of a start up, solid intellectual property protection was the only way to defend the margins of these businesses and create an attractive return for the investors. As the technology infrastructure business matures, however, the nature of defensibility changes.
We do not expect IT enabled services businesses to have strong patent positions that would prevent anyone from entering a market. Instead, we look for defensibility in data.
We would argue that the network effect that makes it difficult to compete with Craigslist or eBay is based on a proprietary data asset. People go to these sites because they have lots of listings. This is a data asset. But it is not their only data asset. The systems for controlling reputation and spam at eBay and Craigslist are also built on a valuable data asset. As these systems accumulate data over time, they increase the value of the service and make it more difficult for newer services to deliver an equivalent experience.
In our portfolio, TACODA, has built a defensible position with a valuable data asset. TACODA works with the publishers in its network to make advertising more relevant to their audiences and more effective for their marketers. They do this by observing anonymous behaviors across the network and then serving relevant ads to users who visit other sites in the network. If TACODA observes a behavior that suggests the user is in the market to buy a car on one site, they can deliver auto advertising messages on another site in their network. TACODA can increase the value of publisher’s audience by delivering an ad that is relevant to their audience, while the publisher can only show an ad that is relevant to the content on the site. This network wide view creates a valuable data asset, but TACODA is also able to see unexpected correlations between behaviors and ad performance (who knew that people with an interest in romantic movies were likely to click through on a car rental offer) and they can also see changes in these correlations over time. Finally, they have had to solve all of the practical problems associated with working with an enormous data set to deliver marketing messages in real time. There is, for example, a point in some campaigns where the cost of managing additional data is not justified by the incremental revenue that could be driven off of those data points. TACODA’s hard won experience with this trade off will take time for a competitor to duplicate.
Information technology enabled services delivered over the internet evolve in a dialogue with their users. They are often created by their users or at least shaped in material ways by them. The data that is created as users interact with these services can be fed back into the services to improve them for the users. We are constantly surprised by the many creative ways this is being done. We do not pretend to know all of the ways that data will be used to improve a users experience and create barriers to entry, but we believe that it is the right place to look for differentiation and defensibility.