Analytics is Unstoppably Reaching All Levels of the Enterprise

Our companies are slow to adopt business analytics solutions in their environment and are thereby losing competitiveness. 

By Miran Varga, originally published in Svet IKT in Delo newspaper on November 7, 2016

Alekander Pivk, PhD., Head of advanced analytical practices (and data management) for the Adriatic region in SAS, says that the domestic enterprises are underweighted when it comes to use of analytics. However, the potential for progress is therefore even greater.

We had a talk with Mr. Pivk about how business analytics is perceived in companies today, what kind of data they need, how quickly analytical tools are developed, in what direction the development of modern analytics is headed and several other topics related to business analytics.

Alekander Pivk, PhD.
Head of advanced analytical practices (and data management) for the Adriatic region in SAS

How do companies perceive business analytics today?

Companies are focusing on the areas that generate the greatest added value and competitive advantage. Business analytics is one of the key factors of making better and faster business decisions that ensure business growth. Introducing analytics at the company level requires a change in culture, processes, knowledge and skills of a large number of employees. Like all major changes the advanced analytics also requires support of management, company leaders have to truly believe in the welfare of quantitative approach. Approximately two-thirds of the companies worldwide already use one or more analytical solutions for making more informed decisions; this share is growing slowly. On the other hand a third of enterprises' are »analytically bare«. In the Adriatic region, which includes Slovenia, there are many more such companies. Enterprises are unfortunately too slow in opting for the use of business analytics in their environment.

Does modern analytics know it all?

Advanced analytics primarily provides sophisticated statistical and mathematical methods and techniques for learning from data or content where the aim is to discover deeper insights, deliver forecasts of events or make recommendations. A simple example is to identify behavioral patterns of people who carry out insurance fraud. Analytic system is delivered a learning set of data first which serves as the basis for distinguishing between good and bad. The better the prediction model the more likely it is the system will be able to separate the fraudsters from the honest people.

There is also an upgrade of this approach where the system learns itself on the principle of rewarding the correct response to a specific input stimulus. The more rewards (or penalties) for a specific response to the stimulus it receives the sooner the system understands the connection. This approach has been used, for example, in the DeepMind system which defeated chess grandmaster Garry Kasparov. On the other hand, it also defines the process of developing such models, from the very definition of the problem, the development model, to the industrialization and efficiency measurement of the final solutions. Analytics itself is not smart on its own but can become very useful and effective if used correctly.

What data do companies need and how do they activate it?

Today organizations have more data at their disposal than ever before, the trend of data growth will only continue. It is true that things are easier said than done; for example to extract meaningful conclusions from the data and converting this knowledge into appropriate action. Companies need to use their own structured data that is stored in transactional systems or (sectoral) data warehouses. Then there is data that is accessible through external providers. Then the unstructured data is entered into equation; for example e-mails, blog posts, data from social media or forums, call center records, all in the form of text, voice, images or even video. By analyzing these we can understand the content of messages, problems, emotion, sentiment, preferences and even age. With the simultaneous use of all this information and suitable analytical approach we can get results which are useful in various levels of the organization. On the one hand companies can optimize the efficiency of agents in the call center, on the other hand they can improve the user experience of customers because they know them a whole lot better.

How do you assess progress in the development of analytical tools - what can they do today that was not possible yesterday?

Development in this area can be viewed from different perspectives. On the one hand you have the advances in hardware performance which is still governed by Moore's Law. Performance of each system is growing, there is the ability to connect systems in clusters, the use of distributed file systems intended for storage and processing of big data - all of these contribute significantly to the fact that we can analyze much larger amounts of data than just a few years ago. On the other hand the development of analytical tools took the exponential curve as new tools and new methods that facilitate and speed up the development and work occur almost daily. There is also a strong open source community and several smaller, narrowly specialized companies with niche analytical tools as well that accompany the larger players who are either on the market for a long time or have in recent years filled their gaps with acquisitions of niche providers. One notable achievement of analytics this year, at least in my opinion, is the fall of the last human stronghold in traditional board games. AlphaGo system has defeated Lee Sedol, one of the three world's top players in the game Go, which is known for its simplicity, but at the same time incredible depth.

What will we be able to do with analytics tomorrow?

Among the most promising areas for advanced analytics are location-targeted mobile advertising, mobile apps, monetization of analytically processed data, several opportunities in the Internet of Things, co-operation of autonomous vehicles, predictive maintenance... The fourth industrial revolution will also bring a number of changes to the social and economic systems that shape our lives and the way we live. For all this, of course, a lot of credit goes to advanced analytics.

Working with analytics requires a range of different skills. Data scientists are highly sought after all over the world. Who is actually a data scientist, what does he/she do and where can companies actually find them?

In recent years the advanced analytics is fast spreading into all branches and levels of organizations. Data science requires a mix of multi-disciplinary skills to achieve the objectives, e.g. better and faster decision-making. Companies seek advanced mathematical and statistical knowledge, knowledge of computer programming and databases, understanding economic trends, legal and business environment. Furthermore a data scientist should still have excellent communication and presentation skills as well as soft skills such as curiosity, creativity, focus and targeting. Data scientists are working on finding solutions in areas that have not yet been studied, where there are a lot of unknowns and the path to solution requires a great deal of inertia, testing and different, unique thinking. People with all these skills are not easy to find. Therefore, at least from my point of view, a data scientist is actually a team of people who interdisciplinary cover all the mentioned areas and overlap with some of the skills. Assembling such a team is anything but easy, but typically it is easier to find people who have the technical knowledge or are good with numbers than those who see the solution or a pattern behind the numbers and are able converted it into tangible and convincing story. 

What is the weakest link of modern analytics and how can it be eliminated?

Actually there are several weak links. In practice many can often be found in the leadership and managers of enterprises who do not trust the modern technology. In our region there is also, unfortunately, the lack of visionaries and an absolute shortage of skilled people. Many people still doubt in progress of analytics as they believe it could jeopardize their jobs and existence, perhaps even life.

Are there any potential pitfalls of use of analytics in business?

Absolutely. Bad, incomplete and misleading data leads to bad results. If you load the system with garbage the end result will also be the garbage. Putting blind trust in analytics can also be classified as a trap sometimes. Processes and decisions should not be based solely on analytical conclusions, the human is still a critical factor in the whole process. The use of analytics creates big differences between companies - analytically stronger companies will survive and prosper as they will much faster adapt to market needs.

Get More Insights

Want more Insights from SAS? Subscribe to our Insights newsletter. Or check back often to get more insights on the topics you care about, including analytics, big data, data management, marketing, and risk & fraud.