Authored by David Pool, UK Managing Director at Curious AI and Corndel Data Advisory Board member.
Data Scientists are excellent problem solvers and business relies on problem solving.
Amazon and Tesla founders Jeff Bezos and Elon Musk, share a common interest in reaching for the stars, but both also have a more grounded passion: making their businesses “frictionless.” For frictionless, read less bureaucratic. Research from London Business School and MLab Research suggests that in the US alone unnecessary bureaucracy adds $3tn to business operating costs annually and globally it is a significant drag on optimisation and innovation too.
Better operational data science is at the top of most companies wish-lists, driven by techniques that can directly impact business efficiency. Organisations are more professional in curating and using data and we now have the ability to model complex processes through advances in statistical modelling, machine learning and deep Artificial Intelligence. Industries and the companies that compete within them, now use data science protocols to create technical solutions to business problems, replacing expert-led bureaucratic systems.
As an example of a data-science led “technocracy”, back in 2000, Investment bank Goldman Sachs employed 600 equity traders in its New York office, handling billions of dollars in daily trading. By 2017 that number had fallen to a mere 4 traders; but they are now backed by a team of 200 data scientists and engineers. In fact, despite being a bank, 40% of Goldmans staff are now data scientists, engineers or computer scientists and the metric cited by the investment banking industry generally, is that 1 scientist can replace 5 traders. The effect on average employee remuneration can be seen in the chart below and between 2011 and 2016 Goldman’s net revenues increased 6% and pre-tax margins increased by a staggering 33.7%.
Data science has dramatically changed the structure of this industry, to the point where an investment bank is really only as good as its data science: namely, automated high-speed algorithmic trading models, the quality, freshness and curation of its data and the speed of its transaction processing networks. For investment banks, this means their data scientists are at the frontline of a technology arms race.
The wider trend is that digitisation allows companies to reshape themselves, focusing a larger proportion of their resources into the reason they exist and less into bureaucratic support functions. In the UK, for example, pre-digitisation, the Central Railway Clearing System employed 1,500 clerks whose role was simply to divide and allocate ticket money between the various railway franchises. Today, the process is fully automated and the central clearing system doesn’t exist.
To be clear, bureaucracy is critical in every well run business, particularly in heavily regulated sectors like financial markets and healthcare, but we can now use data science, AI and analytics to make these necessary processes run as effectively and seamlessly as possible, without slowing down decision making or stifling innovation.
The effect is that most organisations in most industries are trending towards technocracies, where leaders have deeper technology backgrounds, decision-making is data and evidence driven and core business functions are deeply rooted in data science. It’s a trend that started with the tech giants and disruptor start-ups, the vast majority of which are founded by individuals with science and technology backgrounds and who, as the graphic below shows, recruit and train staff with solid data and computer science awareness. Beyond this, Amazon, Apple, Facebook and Netflix all pump millions of dollars into the American school system to promote the teaching of data science, whilst China now has technology schools which teach advanced AI and data science modelling to its pupils, in order to create the future workforce the economy needs.
Every industry is playing catch-up, becoming data driven and more technocratic. One of the main reasons is that, as shown, data savvy technocracies now dominate the world.
In 2007 none of the world’s 10 largest companies was a “data-driven” business, but by 2017 this grew to 7 out of 10. Microsoft acquired over 100 data businesses in this period, including LinkedIn in 2016. It led to a re-invention of Microsoft, from a desktop and office software business to an online digital technocracy. This is now playing out across all industries, with the mantra “the harder a company’s data works, the higher the valuation.” A company’s data strength and the quality of the algorithmic models it uses to service this data can now be used in its overall valuation.
So, now in 2020 data science is equally important to a business as its finance, sales and marketing function. Businesses have clear objectives on what they need to achieve with their data – classifying, clustering, making accurate predictions, running complex simulations, etc and (as below) we now have models to achieve these objectives.
But this is only part of the gain. Data Scientists are excellent problem solvers and business relies on problem solving. The techniques needed to classify data and make accurate predictions in one area of a business can be easily transferred to another, without the need for an expert bureaucracy. So, the real gain is that well-trained and objective led data science teams have impact across the business. This is the model used by the world’s most successful technocratic organisations, which use a project approach to operational leadership, empowered through data science. The role of individual data scientists is to support and sometimes fully automate this process and surface the insight and productivity gains it produces.
As Yann LeCun, Professor of Artificial Intelligence and Director of AI research at Facebook says, “Most of the knowledge in the world in the future will be created by machines and will reside within machines.” For machine, think any system that creates data: cars, business processes, mobile phones, apps, etc. It is the role of the data scientist to speak the data language of these systems, train them effectively, understand how they think and mentor them to be productive problem solvers too. A powerful combination.