In January 2006, Thomas Davenport published “Competing on Analytics” in Harvard Business Review. One year later, the book was released. In September of the same year, Tim O’Reilly published What is Web 2.0?, in which he asserts that “data is the next Intel inside.” These conceptualizations were both a chronicle and a trigger of what has been called the analytics revolution or the data revolution.
During that same period, we witnessed an incredible increase in the quantity, quality and variety of data at the disposal of companies and organizations. Eric Schmidt, former CEO of Google, portrayed this acceleration eloquently: “Every two days we create as much information as we did up to 2003” – an assertion corroborated by the UC Berkeley SIMS study “How Much Information? 2003“.
This shift was mostly due to the generalized use of the Internet for transactions and social interaction, but also mobile devices and the widespread use of sensors. Both the quantity and the quality of data and its attributes increased greatly. E-commerce companies, for example, were able not only to gather transactional data but also to “see” what customers were looking at and not buying – an indispensable analytical tool for studying the intentions and motivations of their customers.
Cloud platforms such as Amazon AWS, Microsoft Azure, Google Cloud and Ali Cloud have been essential in this evolution, not only because they transformed capital costs – the cost of building, upgrading and staffing a data center – into variable costs where companies only pay for the computing power that they use, but also because they transformed the companies’ own business models.
Cloud platforms carried the promise of zero marginal cost (the cost of serving an additional client), together with infinite and almost immediate scalability – a new economics no longer restricted to a few companies, but available to anyone. Indeed, we have all become familiar with these new business models: Netflix and Dropbox, for example, are essentially implementations of the Amazon AWS S3 storage solution.
Even if the more sophisticated computing power needed to train highly complex deep-learning models is now a commodity, platforms – unlike telecommunications companies – managed to escape the curse of commoditization. Telecommunications companies are rapidly becoming providers of IP packets, while cloud platforms are managing to add valuable services and democratizing highly sophisticated artificial intelligence.
The extraordinary success of platforms such as Facebook, Salesforce, Amazon and Instagram offer a preview of this new reality. These were the first examples of the enormous power of these new business models that fully enjoy the benefits of zero marginal cost, infinite scalability, no capital costs and instant deployment.
This success, however, is just the tip of the iceberg. In fact, we have been moving business routines and procedures to software for a long time. First, there were aggregation and accounting processes. Later, online systems – particularly on the web – moved input processes and transactions to customers themselves. This silent march did not distinguish between simple procedures and smart ones. Recommender systems such as those used by Amazon and Netflix to suggest books and movies have been online for almost 20 years and encapsulate a great deal of today’s artificial intelligence. The same is true in operations and in the supply chain, which have been completely transformed by robots. Automated software robots are also taking charge of tax, finance and legal departments – first activated and supervised by humans, and now, in many cases, fully unsupervised.
The first turning point in this process came in the late 1990s with the advent of high-performance and automated trading, which now accounts for more than 75% of all stock trades. We are witnessing a similar process in pricing, where pricing decisions are being transferred to algorithms with no human intervention.
A second turning point came when not only individual functions but a whole set of processes were moved to software. Again, platforms first moved the coordination between buyers and sellers – previously done by sales teams – to software. This was an important milestone because markets are institutions that implement coordination mechanisms. Moving them to software created frictionless markets endowed with total scalability, zero marginal cost, no capital costs and no decreasing returns on scale. These companies not only became the world’s most valuable organizations but did so with very small headcounts. However, important as they are, they are only the canary in the coal mine. Intelligent systems now allow us to endow them with even more sophisticated mechanisms, creating exchanges that were unimaginable just a few years ago.
Moving analytics online was another step in this process of digital transformation. In many companies, analytics is no longer a manual process based on static datasets; instead, it is completely embedded in code and in the cloud, integrated with the rest of the organization’s processes.
James March, best known for his research on organizations, coined the term exploration-exploitation to describe the organizational tension between running the current business and exploring the new. Now that exploitation is largely done by code, companies have become exploration devices that compete in innovation with a new set of economics and are endowed with both human and non-human logics executed at the speed of light, at zero marginal cost and with infinite scalability. Companies are becoming code. This is a new reality that demands not only different capabilities but also a new look at competition and regulation.