From waking up to LINE messages, asking Google Maps for the fastest route to work, to consulting a calorie-counting app on whether we deserve a sumptuous dessert at dinner, we leave massive digital footprints with everything we do, eat, buy or even desire.
With the exploding rate at which digitised information is collected, corporates are racing to make sense of data and extracting actionable business insights and opportunities. No wonder we are seeing a growing trend among businesses to build data analytics teams.
The question is then why build, not buy from big data startups cropping up everywhere.
Sutapa Amornvivat, Ph.D. is Chief Economist and First Executive Vice President at Siam Commercial Bank. She has international work experience at IMF, ING Group and Booz, Allen, Hamilton. She received a BA from Harvard and a PhD from MIT. Email eic@scb.co.th. EIC Online.
To be sure, buying services from vendors is another way to obtain analytics capability. Yet, there are clear benefits of building your own team that are irreplaceable over buying readily available solutions off the shelf.
To fully utilise analytics capability, a high level of engagement is crucial. Machine learning algorithms, a key ingredient in data science, needs time to let machine "learn" before it can truly be of any use. A hired helper may not stick around long enough to see through this learning process, undermining the potential of the powerful technology.
Moreover, even though rolling out is always the end goal, most vendors can only promise performance, well-tested on delivery. Adoption of the technology at all levels from top generals to foot soldiers will require more persuasion and commitment. The change of culture toward a data-driven organisation can only be instilled from within.
Over time, building your data analytics capability will get easier and faster as machine learning technology becomes more and more commoditised. Take IBM Watson, a supercomputer known for beating champions of Jeopardy! TV show in answering quizzes. It has evolved into a customisable cognitive system, or an artificial brain that can be trained to answer specific business questions, and is now being offered as a service by IBM. Amazon Web Services and Microsoft Azure are providing machine learning capability in their cloud services as well. This means almost any business will be able to engage its customers better through data intelligence. A serious downside of building your own analytics team is cost, because the talent pool is small.
At the centre of it all are data scientists -- a group of professionals with technical skills to find patterns in complex data. Harvard Business Review calls it the sexiest job of the 21st century. Translation: costliest.
Glassdoor, a job search engine website, reported that job openings related to analytics on its page surged by 147% from 2016. The site also ranks data scientist as the best job to have now in the US, based from available job openings, salary and job satisfaction ratings. LinkedIn says adding "data science" as keywords on its job networking site can fetch three times more views than most of its 500 million profiles!
Data scientists are a relatively new profession. They can come from different disciplinary backgrounds such as computer science, physics, mathematics, and engineering, with a unifying theme of strong quantitative skills and ability to write computer codes.
However, as corporates compete for this new breed of talent, they should not dismiss other roles required to deliver a real business impact. Even the smartest algorithms will not yield a fruitful result, unless they are put into use. Adoption is the key to success in an analytics transformation.
In tackling a real-world problem, a well-balanced team is required. It is crucial to develop a deep understanding of the business context for the problem at hand. This can be the task of experts in the domain who also have a good grasp of data science approaches. They must understand requirements and pain points of end-users necessary for designing analytics solution and devising a business strategy for its adoption.
Another key member that can greatly improve efficiency for the team is data engineer. A large part of the data science process is about data management and preparation. Data engineers can handle this difficult and the most time-consuming step allowing data scientists to focus on developing complex algorithms.
Finally, in creating business impact, analytics results should be packed in a deliverable format. Different business problems call on different forms of solutions ranging from a few lines of business insights to a compact mobile application to a full-blown software suite. For this, software engineers can add great value. In fact, this can make or break the outcome. Be it five or 50 members, these are the core units that define an analytics team. Effective dynamics is a prerequisite to unlock a firm's advanced data capability.
As the world becomes increasingly digitised, new data will trickle in at a breakneck speed. The game will no longer be about who owns more data; it is rather about who has the power to distinguish meaningful patterns from noise. One misstep can make you fall behind the competition, if not simply game over.
Thus, setting up a wrong team to harness that power is undoubtedly worse than not having a team at all.
