Recent excellent articles by Martin Casado at Andreesen Horowitz and Reza Zadeh at Stanford make an argument that data moats are a myth. This stands in contrast to arguments made by people like Ivy Nguyen and Jocelyn Goldfein, formerly of Facebook and currently a partner at Zetta Ventures. Really, the two are not so far apart, data moats do exist and are important. But having access to data alone does not create them.
A data moat requires two essential things, and missing either one of them means your moat is not really a moat. A data moat needs 1) an accurate collection of relevant data, 2) the culture to effectively use it. Having years of experience working with and consulting at enterprise e-commerce companies, I am consistently surprised by how often the latter is missing and how this turns a would-be moat into a puddle.
There is a point made by detractors of data moats that data alone rarely builds great companies and that the world is littered with examples of small start-ups with far less data bringing down behemoths with access to many times more data. This is absolutely true, but it stands in stark contrast to companies like Google, Amazon, and Yelp.
Google won its position through creating and consistently innovating on the best website search algorithm in the world built on its mountains of data, and using the algorithms it developed on that data to destroy its competition through its more relevant search results. Amazon was the first retailer to effectively use its data to create completely new product discovery experiences like recommendations and to allow for effective and relevant searches over its massive product catalog. Yelp may be the best example of a data moat. It captured more restaurant review data early on than anyone else and no one has been able to effectively compete with this data advantage for most of the life of the company. Of course, all three originally had far less data than their competitors, but they were able to more effectively use the data they did have and eventually turned that effective use of data, coupled with their eventual collection of massive quantities of data, into an effective moat.
The reasons Google, Amazon, and Yelp were able to create effective data moats, but brick and mortar retailers who originally had far more data were not, is something I continue to see first hand. A company like Toys R’ Us had massive amounts of data on its customers. It could have used that data to personalize those customers’ experiences in ways no one else could. It could have used that data to learn about where their customers were having bad experiences and fix them. It could have used that data to give its customers the best product discovery experiences available anywhere. It didn’t. And it wasn’t for a lack of data.
Instead, the problems for most older enterprises tend to be rooted in cultures that pay lip service to data’s effectiveness, but are not able to turn that into prioritizing the effective use of data. I have personally witnessed multiple examples of companies I’ve worked with who ran A/B tests without outlier detection, or collected data, built products on that data, but lacked controls to verify the data was accurate (it wasn’t). These are the types of problems that take what could be a data moat and drain it. They are also the types of problems that tend to be endemic to a culture and are incredibly difficult to change from the inside. The engineers at these companies who try to change them tend to eventually get fed up and leave. It is a hard thing to change a culture, but without the right culture, data is useless.
Once Amazon showed the power of data and technology, its competitors raced to catch up by forming divisions aimed at operationalizing their own data, but aimed to do this internally. At the time, they still had more capital and more data than Amazon, but arguably Walmart, with its 43% year-over-year ecommerce sales growth rateis the only one really succeeding, through aggressive acquisitions and third party partnerships that forced a more data-centric mindset into the company from the outside and made it rival Amazon in technological capabilities.
In the end, data moats do exist and some of the most successful companies in the world effectively use them, but they are not predicated on data alone. Without a data-centric culture that ruthlessly prioritizes the effective use of data, no moat can exist. But, companies who have created such a mentality and coupled it with an overwhelming preponderance of collected data, have been able to use it as a massive competitive advantage. This is both a warning call to enterprises with massive amounts of data, but the wrong culture to use it, and a rallying cry to startups who can create a data-centric culture, gather enough data to make it useful, and beat their larger competitors to creating a moat. Data moats are real, but data-centric culture is as important as the data itself.