Martin Casado and Peter Lauten, both Partners at the well-known venture capital fund Andreessen Horowitz (with investments in Facebook, Twitter, Groupon, Airbnb…), recently published a post called “The Empty Promise of Data Moats” on their fund´s website. Within this post, they present a framework aiming to help understanding the data journey of a start-up building up its business and database.
They start with the “chicken-egg” problem, meaning that in a business requiring a network it can be difficult to acquire the first customers or nodes. Yet, they point out that for a business with data scale effects, the initial data required for first training (of an AI) can also be gathered by data captured from available sources such as the web. This shows that the data set required for a minimum viable product (MVP) is not enough to defend a business, as it often contains relatively cheap and publicly available data.
In the next step, Casado and Lauten point out that data acquisition costs, as opposed to traditional network effects, go up over time. This is explained by an increase in difficulty to find unique data that improves the model, as much of the newly acquired data overlaps with the existing set, which also makes the data less valuable.
As a last point, the venture capitalists mention the need of data freshness. This refers to changes in the underlying real-world state, which does not stand still. For instance, a fictive algorithm designed in the 1990s with the purpose to help users chose which clothes to wear each day would need a high amount of work and fresh data to not propose outdated outfits.
When looking at these points, one might wonder why it often seems that all value and defensibility was inherent to data. Although data plays a significant role in a company´s success, one should not think that having a lot of data translates into having long-term defensibility. Rather, a company needs to add specific strategies that allow defensibility when paired with data. These strategies are for instance the securing of proprietary data sources, understanding which data improves the core product or strategies that keep the data freshness up. Therefore, data alone are not the key to defend a business from competitors moving into its space.
Bibliography:
Martin Casado and Peter Lauten 2019, The Empty Promise of Data Moats, Andreessen Horowitz, viewed on 14 September 2019, https://a16z.com/2019/05/09/data-network-effects-moats/
Very refreshing post Ivo!
As you pointed out in most of the news articles these days and even our study program only talk about data as the lone source for a firm to create value or maintain its competitive advantage. However you mentioned some very eye-opening facts:
The crucial part is not owning data but using this data to gain the correct information and then pairing this with the most feasible strategies. Also, bigger is not always better. There is no advantage of owning a vast amount of historical data when your data outdates very quickly. The aspect of data freshness gets overlooked too often nowadays.
But the part that sparked my interest the most was the fact that start ups can acquire their data through the web. This made me think about different methods Start Ups might use to gain more data. One of the methods I could think of is a crowdsourcing strategy calles “user-in-the-loop”, which involves designing side-products that gives customers the incentive to provide additional data to the overall system (e. g. Facebook’s photo tagging option). Another method that pops into my mind is also creating side businesses that are offered for free and targets existing consumers. This is very popular in computer vision Start Ups, such as Clarifai, offering free photo apps providing them with data for their core business.
There are many options more for Start Ups to increase their data volume so this is clearly a very interesting topic providing a lot of room for thought!