In today’s digital age, data is king. With the rise of big data and advanced analytics, businesses are able to gain insights into their customers, operations, and markets like never before. At the same time, with the growing importance of data science, more and more companies are hiring data scientists to help them make sense of this flood of information.
While data scientists are experts at analyzing data and building models, they often lack the business skills needed to translate their findings into actionable insights.
This highlights the importance of data scientists having a deep understanding of the businesses they work for and the industries they operate in, while they also need to be able to understand the problems their companies are trying to solve and how their work can contribute to achieving business objectives.
Precisely this challenge was the starting point for Croatian entrepreneur and scientist Sinisa Slijepcevic and his endeavor in data science. With a PhD in applied mathematics, Slijepcevic’s experience includes working for global consultancy companies such as McKinsey. It was there where he recognized an existing gap in big data’s relationship with the business world.
“There was a huge gap between what you can do with data and what businesses really need. These were two worlds which can’t speak with each other. On the other hand, business leaders know that they need to leverage data, and also need to understand how to use huge new capabilities, because they don’t really know what you can actually achieve with it. And they don’t seem to speak the same languages – they think they understand each other and they speak very often, but there are so many failures because of this,” Slijepcevic tells The Recursive.
Understanding business specifics and ML technologies
Bridging this gap then led Slijepcevic to founding his startup Cantab Predictive Intelligence (Pi), whose solutions combine deep practical understanding of both business specifics and the most advanced machine learning technologies.
The company’s main offering is a machine learning as a service (MLaaS) through their own cloud AI platform, focused on the financial and pharmaceutical industries in areas such as credit rating calculations, sales optimization, and communication optimization among others. Their clients also include some of the biggest banks across Europe and Africa, such as South African Nedbank.
For example, one of the approaches that Cantab Pi’s uses is to calculate AI engagement predictors by leveraging more than 5,000 daily updated dynamic data points, which are then aggregated and maintained in a data model. The client for this approach was a big pharmaceutical company, and the results showed a 3 percent higher sales after four months of deployment.
One of the pillars of Cantab Pi’s success when working with these big industries is that all of the data scientists in the team need to understand business, Slijepcevic explains.
“It’s difficult for them since it’s very far out of their comfort zone. But we insisted that they need to understand the business problem and they need to understand every single data point we use and how this data was created. What is the workflow which created this data point? If you don’t have this understanding, then within a couple of minutes or maybe a couple of seconds, this data scientist will make a wrong decision which will result with days, months, and years of work becoming completely useless,” Slijepcevic tells The Recursive.
According to the Croatian founder, there are different data points that data scientists need to pay attention to and which are then used by clients for their sales.
“Each sales representative gets a recommendation for which channel, which message, timing – they get precise instructions for whom they need to visit, how, which email they need to send, and so on. All of this data is calculated by our platforms and then sent to the companies. And when they do it, the companies that are deploying this earn 20 to 30 percent more compared to when they don’t,” he explains.
Several skills that can help data scientists become more business-savvy
Few business skills that data scientists should also have is to know how to prioritize achieving the maximum return on investment while minimizing the use of time and resources, rather than emphasizing the complexity and grandeur of their code.
They should start with a basic approach and gradually expand their framework as the project proves its worth. Additionally, data scientists should be able to recognize when a project is no longer relevant and abandon it promptly, an area where persistent data scientists may struggle.
Real-world examples also showcase the importance of business acumen for data scientists coming from the healthcare industry. The healthcare sector is generating vast amounts of data from electronic medical records, clinical trials, and wearable devices – and data scientists here need to be able to understand the needs of healthcare providers and patients, as well as the regulatory environment they operate in.
For example, data scientists can use machine learning algorithms to analyze patient data to identify high-risk patients and develop personalized treatment plans. However, they also need to be aware of the ethical considerations around using patient data and the potential impact on patient outcomes.
Another example comes from the e-commerce industry, where data scientists are using machine learning algorithms to develop recommendation engines. These engines use data on customer behavior and preferences to suggest products that customers are likely to buy.
However, data scientists also need to be aware of the broader business objectives of their company. For example, recommending products that are not profitable for the business may lead to short-term gains in customer satisfaction but harm the company’s long-term financial performance.
According to Slijepcevic, such examples just illustrate the growing need for solutions in the particular niche, additionally motivating data scientists to become more business-savvy. An approach that the company is taking is to first hear what are the main needs of the companies, and then offer a sufficient model.
“Our offering is based on the feedback that we get every single day from our customers. So we always start with a deployment, and not a concept or brainstorming or something else – and this approach helps us stay a few steps ahead of what anybody else is doing right now,” he adds.
At the same time, each particular use case requires a different approach and this is something that will become even more common in the upcoming period – another skill that data scientists will need to acquire when it comes to understanding specific business needs.
“I think that there are going to be tons of use cases where you will have very specific data architecture, machine learning architecture, and so on, which is going to revolutionize this particular area – it will be a blend of business statistics and machine learning statistics models tailored to particular use cases.” Slijepcevic concludes.