Rethinking information management
Video | Infographic | Blog Post | Trend Cards | Tweet Chat
The boundaries of the data universe are expanding to accommodate information of every type, format and source known to man and machine, which according to one estimate amounts to 3.6 zettabytes. From “some” structured data, our focus is shifting steadily to “all” data – structured and unstructured. And yet a humongous segment of this unstructured data – rather vividly labeled dark data – goes unanalyzed. In just the last year and a half, almost two-thirds of enterprise leaders I’ve talked to said that they had more than a few unstructured data stores lying under-analyzed or even unanalyzed, and more than a quarter said they had several. A growing desire among companies to understand big data is driving them to take a fresh look at their data and ways in which they store, manage and analyze it. This is indicative of a progression in the big data evolution from “what and why” to “how and now.”
The ascendancy of big data in organizational priority will now force a rethink of enterprises’ information management strategies, with its early beginnings unfolding before our eyes in 2014.
Data analytics will evolve from establishing relations between conventional data to establishing new and unexpected correlations
A bigger change than the “I want to store it all” syndrome will be in the way that organizations use data to make predictions. Traditional data analysis seeks to establish direct relationships between different pieces of data – in other words, it tries to track the cause and influencers of a particular event. In contrast, with big data and its expanding scope, there is ample opportunity to uncover correlations – data that may seem unconnected, but relate to events that are connected – enabling surprisingly accurate, or sometimes just surprising, predictions that give rise to new business imperatives and ensuing decisions. For instance, urban legend has it that the sales of beer and diapers move in unison. The correlation goes like this: diaper emergencies usually occur in the evenings or later, at which time the father is dispatched to get fresh supplies, which is when he also ends up buying his beer!
Big data will enable organizations to uncover hidden insights about their business, and it will also enable them to take appropriate action. This was almost impossible earlier when organizations had only limited, structured data to play with. With the widespread realization that all data needs to be scoped within a framework of holistic information management, enterprises will seek to tackle the challenge of “how’’ that surrounds the best possible applications of big data.
Analytics will be recast in a do-it-yourself “uncover and respond” model
The bid to evolve from understanding just ‘data relations’ to exploring ‘co-relations’, will begin to transform the enterprise’s predictive model-driven approach to data analytics. Now business users will be enabled to first discover and find correlations between data. And then also respond to it with appropriate timely action. This is more challenging that it sounds because correlations are usually discovered by trial and error methods and sometimes even by accident. So the new analytics model will have the flexibility and scale to try out a large number of permutations of different sets of data to support self-discovery of correlations. Then, it will also have the ability to launch appropriate actions in a self-serve mode.
By working on relationships and correlations between data, predictive modeling and analytics, enterprises can improve the quality of, as well as the time to, insight. This means that big data will actually enable faster, better, more fruitful informed decisions.
Big data will spell the end of master data management, for analytics
In years past, organizations aimed to create that one single source of truth, which would house their master data. However, for reasons of expedience these organizations started to replicate their “master data” sources, defeating that very purpose. That was then, when data was limited and fairly well structured. With the scope of data changing to all data, it is virtually impossible to sustain a pool of clean master data in an ocean of facts, figures, numbers, records and conversation logs. Indeed, this may be not even be required. Moreover, with the construct of algorithm-based analytics, and the maturity of evolving algorithms, the need for master data management will soon be redundant.
In fact, while master data management seeks to restrict data in terms of form, format, veracity and so on; big data explodes these limitations to encompass just about any stimulus, which carries information. This strikes at the very foundations of master data for analytics.
Data will increasingly come from machines and things
Think of the currently estimated 7 billion connected devices in the world. Organizations will start to leverage M2M (machine-to-machine) data and from the IoT (Internet of Things) for product innovation, efficiency improvement, preventive maintenance and so on. That’s not all. In a reversed order of things, machines talking to other machines will share information, spot correlations, draw conclusions, and only then serve them up to human beings for further action. This will be an unmistakable sign of the impact of big data on information management.
Futuristic? As early as 2014.