“The sexiest job of the 21st century,” as the Harvard Business Review puts it, data scientists are fast becoming one of the key roles in modern businesses. With companies generating unprecedented amounts of data on their customers, it’s become more and more important to hire specialist skills to sift through and interpret that data.
However, problems arise when businesses around the world wake up to the necessity for data scientists at the same time. McKinsey, the consulting firm, estimates that there is a global shortage of data scientists, with the US alone suffering a shortfall of 140,000 to 190,000 people with ‘deep analytical skills’, as well as a shocking 1.5 million data analysts and managers.
This obvious supply issue is masking a problem far more intrinsic to the role of the data scientist. Even companies that do hire data scientists often find that they know comparatively little about how to leverage their skillsets. As the MIT Sloan Management Review puts it, many businesses with data scientists “struggle to realise the full organisational and financial benefits from investing in data analytics”.
There are several issues at play here which are preventing the data scientist role from realising its true potential. Data scientists suffer from often being managed by people who do not understand how best to leverage their specialist skill sets. If managed poorly, these employees can become isolated from the day-to-day running of the business, unable to communicate their insights and build the solid business intelligence required to create a great customer experience. Let’s be clear: this isn’t a problem with the data scientists themselves. Rather, it’s a structural issue of the businesses that hire full time data scientists.
Ecommerce companies such as Asos, Zalando and boohoo.com have led the way in using data science to transform their businesses: without the ability to physically see their customers’ behaviour in brick-and-mortar stores, these companies were among the first to realise that data analysis would form a key part of their customer experience strategy. Metrics like cart abandonment rate, ‘broken journey’ and even mouse movements can give significant insights into what’s working and what’s not.
McKinsey, the consulting firm, has found that retailers using big data to the fullest could increase operating margins by more than 60 percent. However despite this huge potential, retailers are still struggling to create true value from their data. For many companies, ongoing data reporting has become routine, something that clogs up data scientists’ inboxes with admin and is rarely read by employees.
Ecommerce companies are very good at generating reports on what is currently selling, or how well the site is currently performing, but they are far less effective at explaining why this is the case. It’s easier than most people think to begin uncovering the ‘why’ – it doesn’t even take a highly-qualified data scientist. What’s required are simple online analytics tools which enable marketers and salespeople to track everything customers are doing on their sites, and develop watertight customer journey mapping. This would allow companies to step beyond knowing what their customers are doing, and move to answering why they are behaving that way.
It’s hard enough for data scientists to communicate their worth to senior management. Often managers know data analysis is important but lack the deeper understanding of what data scientists actually do. What they commonly see is an employee being mismanaged into acting as a gatekeeper between basic company data and the wider business. This data should be readily accessible to all, and data scientists need to focus on what it is they do best: identifying broader trends and converting that insight into actionable business intelligence.
The way to fix this is deceptively simple: we need to shrink the role of the data scientist. The intermediary between website performance data and employees should be delegated to an automated tool. This information simply doesn’t need as much human time and attention as businesses are currently giving it. There are many advanced visualisation tools on the market which can translate numerical data into qualitative insights, without having to use up the time of an expensive data scientist. We should be leaving it to automated tools to routinely deliver insight on how customers are behaving, and why they are behaving that way, leaving more complex ‘one-off’ tasks to a data scientist.
For example, interactive heatmaps from aggregated mouse-tracking superimposed over product pages can be incredibly educational to a company’s ‘rank and file’, demonstrating insight into customer experience in an easily understandable visual language. This type of communication is sorely needed. We need to move information about how customers are using products closer to those who are best situated to do something about it, the product managers.
Businesses, both those in ecommerce and beyond, are continuing to change the way they approach analysis of their own data, and the role of the data scientist needs to evolve to reflect that. We need to recognise that some of the activities data scientists are currently burdened with could be handled by automated toolsets, and that by offloading that insight, businesses would allow data scientists to play to their strengths. Refocus your expensive, highly qualified data scientists on developing real business intelligence, and at the same time foster a culture of data-awareness and data-led decision-making among your wider employees by using tools that make data understandable and accessible.
Duncan Keene is UK managing director at ContentSquare