Deriving ‘Big’ Value Out of Big Data

Until last year Facebook saw over 2.5 bn content items being shared every day. Twitter now sees over 200 mn active users, sending over 400 mn tweets per day. Just half a decade back, in 2008, Google was processing over 20 petabytes of data a day, while the total internet traffic was just 100TB, for the year 1993.

According to the forecasts by IDC, humans will be generating over 40 zetabytes of data by 2020. In India alone, the digital universe is expected to grow from 127 exabytes to over 2.9 Zetabytes of data by 2020-a staggering 2300% growth.

This unprecedented growth in data over the last two decades and its continued up surge can be attributed to the ever growing Internet traffic, smartphone penetration -beyond comparison-and increase in online social networking platforms. This, along with machine generated data, such as call detail records, network events, web logs, RFID information and telemetry data has resulted in the phenomenal growth of digital data.

This proliferation of data has made big data-often typified by the volume, variety and velocity of data the game changer. Exploiting big data solutions can help in various ways, such as grabbing market opportunities, reducing customer churn, building better and innovative products, reducing operational costs, managing risks with prudence and improving financial outcomes.

However, big data is more than employing cutting edge tools and technologies, applying complex machine learning algorithms and exploiting visualization techniques. In order to enable the institutionalization of (Big Data) analytics, the right man-machine combination of people, processes, tools and technology platforms will have to synergize.

Defining the business objectives and context

Most companies, while taking into analytics initiatives, fail to define the very business problem with clarity.

For instance, a retailer may want to develop a strategy to reduce customer decline, or a pharmaceutical company might want to maximize marketing program effectiveness, while minimizing the cannibalization of prescriptions. Such business problems start off as muddy and fuzzy and require multiple functional groups within the organization to engage and deep dive into the problem to uncover the context and make it clear.

It is only when the business problem becomes clear that a robust solution, enabled through analytics & technology, can be devised.

Companies thus need a structured approach to define, articulate and represent such business problems which can only be achieved when multiple stakeholders, within an enterprise, collaborate to scrutinize the business problem. This also helps to identify and fill gaps in terms of unavailable data, for which necessary assumptions can be taken. The collaboration would also ensure that underlying problem is uncovered, instead of just the symptom.

Conclusively, while taking on big data initiatives, companies should ensure that it does not end up becoming a ‘Technology' only project. There should be a constant handshake between IT and business teams to harness real benefits of big data.

Interdisciplinary approach for analytics adoption

The ever changing landscape of business environment has resulted in addressing constantly shifting and ill-defined business problems. In addition - data deluge, increasing technological complexities and advancements in application(s) present itself as a mixed bag of business problems, which organizations have to deal with on a continual basis.

Big data analytics thus calls for an interdisciplinary approach. This encompasses having the right business understanding (knowledge of business domains, such as marketing, risk and supply chain and verticals such as retail, financial services, pharmaceuticals, technology, etc.), an ability to apply multiple math disciplines such as statistics, econometrics, etc. And use of enabling technologies that includes BI platforms, visualization & statistical tools, among others.

The other important ingredient is behavioral sciences-an understanding of how organizations and humans absorb and process insights. This will help to develop nudges (cognitive repairs) against possible human biases while taking business decisions.

The competitive differentiator is in the consumption of analytics

Data is becoming increasingly granular and pervasive, while the demand is increasing for real-time analytics (of various levels and types) by the customers. Companies are willing to invest significantly on procuring cutting edge business intelligence tools with intuitive and engaging interfaces such as interactive dash- boards, with heightened ‘appealing quotient' on presenting the insights. This becomes futile until the focus is equally given to the consumption of analytics.

More so, it is an iterative and recurring process that encompasses communication & implementation of analytics, its measurement (through visible RoI), aligning necessary incentives to promote a culture of data-driven decision making and finally developing cognitive repairs to let the facts rule the process.

Companies willing to reinforce a culture of analytics and data-driven decision making should rather be greedy about analytics consumption. As the domain of data analytics mature, we will see more and more companies competing on their analytics consumption capabilities.

Technology enablers for consumption of analytics

Data visualization & usability is one of the key macro trends that underpin analytics consumption. New, intuitive and engaging interface such as interactive dash- boards, decision-support application simulation tools, etc. and visualization technologies such as HTML5 & extended JavaScript library including D3, jQuery, Raphael, WebGL can enhance visualization capabilities enabling better analytics consumption.

Besides, the data today is not static and is always on the fly and the latency between data collection and decision making deters effective analytics consumption. It becomes imperative to design & deploy or invest in capabilities that harness the available computational power and utilize data optimally.

Remarkably, the next generation business applications are not only to align with transactional execution processes but to literally drive the execution. This can be accomplished through intelligent systems.

By definition, an intelligent system can display properties of intelligence such as perception, memory, correlation, inference, anticipation, reaction, communication and retrospection.

As these ‘intelligent systems' mature, they will abstract the user (analysts) away from the data and empower them to focus more on consuming analytics and driving data-driven decisions.

Cross-pollination of ideas and convergence for problem solving
One of the most innovative ways to solve complex and a variety of business problems is to unbundle knowledge and learning opportunities by expanding the domains of best practices across industries.

This approach of seeking wisdom of problem solving from one's industry and its application on others can offer startling business benefits when applied to big data analytics.

For example, online advertising networks can learn how to manage ad space by learning how airlines manage seats; a health care organizations could become more efficient by learning from the assembly chain operation in a manufacturing industry; or retail and financial services can learn about price optimization from hospitality and tourism industries.

Conclusion

The significant reduction in data storing costs and phenomenal increase in computing power, and other emerging technology trends such as cloud computing, has accelerated the growth of big data solutions. It has been established that the dominant predictor for corporate success is how well companies use their data. This makes big data more than a buzz word or a catch phrase. The competitive differentiator comes with the unique combination of data-set, skill-sets, tool sets and importantly, the right mindset. In simplistic terms, it implies that the individual components such as people, processes, tools, and technology platforms must come together as bionics.

But because the investment of time, money and infrastructure of undertaking such analytics initiative is high-companies need to set the objectives and plans clearly and consciously before delving into it. It is also important to create a robust analytical governance model, inside the organization, that ensures cross-organization collaboration.