It may sound simple but the primary reason most organisations fail to derive value from big data is that the right people cannot see and interact with it in one place. Backend data scientists play an important role but ultimately they’re not the ones making the final decisions in relation to key business problems.
True decision support can only be achieved by putting all of the relevant information in front of the senior management teams and process owners at the point that actions are being discussed and agreed.
For example, we’ve probably all sat in one of those meetings in which we expected to arrive at a clear actionable outcome, only to find that the data we are evaluating prompt more questions than they answer. The very precise questions we have asked have resulted in very precise answers.
However, when we see the resulting outputs we suddenly realise there are other equally important factors we haven’t considered or accounted for. What follows is a request for further analysis which delays the reaching of an actionable outcome. In the worst cases it leads to an organisational stupor in which the status quo is retained simply because doing anything new is too difficult.
Business leaders tasked with seeing and acting based on the big picture are too often forced to do so based on a narrow view of the business which frequently leads to sub-optimal long-term decisions.
> See also: How to sail the perfect storm of big data
For example, looking at a single performance measure such as sales data in isolation could obscure the significant impact of variables such as currency fluctuations, supply chain problems or even disparities in staffing levels.
In my experience most organisations have already made multiple investments in their back-end big data management systems but the problem is that their front end business intelligence viewers are too narrow.
The viewer acts like a magnifying glass when what they really require is a telescope. For example I know of one company that had a supply chain process built upon 14 different systems. Each system worked perfectly well on its own terms, but it was impossible to display the outputs in such a way that they could be combined on one screen.
Most software-based enterprise viewers are designed exclusively to enable the user to dig deeper and deeper into the data under analysis to enable the user to gain ever more granular levels of insight. This does sometimes have its place.
However, if what the user really wants to do is to combine and view multiple sources of data then it requires significant levels of customisation to ensure the all of the systems can interface. It forces them to simplify all of their information to its lowest common denominator and often to strip away the very essence of the data which made it worthy of analysis in the first place.
In such a scenario what the user really needs is the ability to display each data source side-by-side without stripping away any of its complexities. In effect to create a virtual 'control room' which displays the entire ecosystem and enables new inputs to be plugged in as and when required.
Only then can they interrogate the information perform 'what-if' analyses and begin to spot the significant patterns. Enabling this has little to do with technical implementation or the functionality of bespoke integration of software. Instead it has everything to do with the size and flexibility of the viewing screen.
> See also: Machine learning set to unlock the power of big data
For CIOs and IT teams it may initially seem counter intuitive to focus on hardware as a solution to the issue of realising value from big data. However, if the back-end processes are already broadly in place then it is the physical and digital applications that enable the presentation and manipulation of the resulting information that becomes vitally important.
To truly act as an enabler for the wider business the IT department must ensure software is more accessible to business users and that enterprise viewers can be configured and operated without their assistance.
That no matter how complex the individual pieces of software, there is no barrier to pulling up various data feeds side-by-side in order to spot the significant patterns.
We can all agree that getting the most out of big data relies on being able to see the big picture. It therefore follows that the desired outcome can only be achieved when we match the right analytical capabilities with the appropriate advanced visualisation tools.
Sourced from Jonathan Priestley, VP, Multitaction
