While data can be used to reduce costs and improve internal processes, its true value is defined by how these businesses collect it, manage it, and use it intuitively to gather insights, says Ashok Suppiah
Data is the differentiator. That’s what you will read in any contemporary business journal. Yet, despite the urgency associated with the term ‘data’ (and more recently, ‘big data’), it can often feel opaque and therefore difficult to formulate into a coherent strategy. Business leaders’ ability to see over the hill and make mission-critical decisions is more crucial than ever, but that hill is now obscured by a glut of digital information and savvy adversaries.
The margins between success and failure are shrinking. Recent decades are littered with stories of businesses that failed to see the digital transformation in their industries coming. The ones that have been successful, however, understand that they must become more adept at leveraging their data if they are to rival their highly agile, data-driven competitors. Market research specialist Forrester estimates that these ‘insights-driven’ companies are poised to wrestle around US$1.2 trillion away from their rivals by 2020.
At a basic level, data can be used to reduce costs and improve internal processes. Its true value, however, is defined by how these businesses collect it, manage it, and use it intuitively to gather insights.
The good news is that the technology is already here. Now, it’s about asking the right questions and, for those who are responsible for making business decisions based on data, learning how to analyse information as quickly as possible.
This begins by understanding the four elements that give data its transformative power: size, format, age, and discovery. By using these categories as a framework, businesses can decipher their digital information in a way that informs better decision-making and allows them to gain that all-important competitive advantage.
Size matters
‘Big data’ simply means lots of it. The sheer scale of today’s networks and online platforms has provided organisations with an immense volume of metrics per device. Gartner’s oft-cited estimate that 20 billion ‘things’ will be connected to the internet by 2020 gives some indication of the scope. But the trend is creating an unassailable challenge in statistical inference. From the myriad data now available it is easy, for example, to infer things that do not hold up in the real world or might not be that interesting to a particular organisation.
Collecting and analysing all this information – let alone determining what’s useful and what isn’t – is impossible with human analysis alone. Even legacy IT systems are unable cope with the velocity, volume and variety (the three Vs) of today’s data sets.
For a growing number of organisations, then, the solution is machine learning. While the technology might seem like another arcane concept, its use and benefits can already be found in many aspects of everyday life, from spam filters on a web browser to emails that offer recommendations or discounts for online retailers. Machine learning tools are simply sophisticated computer programs that use algorithms to perform increasingly accurate analyses of the data that they collect. One of the best examples of this is Google’s AlphaGo system, a computer program that recently beat the world’s best player at turn-based strategy game Go. Google’s system works by analysing previous moves and combinations – which can amount to millions – effectively making it a stronger player with each new game. The parallels in commerce are easy to identify. For online retailers, each new piece of data allows them to build a more accurate picture of their customers’ buying habits.
Form follows function
Contemporary data can be grouped into one of two categories: structured or unstructured. Structured data is info conventional information that is clearly defined and easily accessible through sophisticated database query languages. A great example is online banking systems that record account transaction information including the date, the amount, a short description such as the source of the money, and payee details.
The world, however, has moved on from structured numeric data to non-structured, non-numeric, multimedia-type data, which includes text, image, video, and animation. A Twitter user will post a combination of these – a photo, gif or video, and an opinion – the latter being a piece of subjective information that would be gathered by performing what tech circles refers to as ‘sentiment analysis’.
Again, online retailers offer a great example of how unstructured data can be utilised to market accurately to each customer. Here is one elaborate yet possible scenario: when a customer enters a retail outlet with their mobile phone, their movement can be tracked. The store might not know who the customer is as that would depend on whether it recognises their mobile number, but it has the ability to record what sections they visit and how much time they spend in each area. Security cameras eventually pair up the customer’s image with their phone so that the store will recognise them in the future.
During their next visit, if the customer were to walk by a particular product and the data indicates that the customer had looked at it previously, the retail outlet could send a text message offering discount. The store could also have a system in place that tracks competitor prices, allowing it to perform a price comparison and offer an adjusted price in real-time. Once the customer purchases an item, the store could then take advantage of further selling opportunities for related products.
Age is but a number
Much of the data now in play is very immediate. In the past, it took months or even years to run a decision analysis project – from collecting the data to cleaning it up, loading it in databases, building models, learning new insights, and finally taking action. By the time a decision had been made, the data would often be stale.
Now, though, the cycle time of analysis has been reduced down to seconds or even real-time, and often to great effect. In the retail store example, obtaining price comparisons once the customer has left the building will be much less effective.
The voyage
While lots of progress has been made in the field of data collection including the process of uploading it to networks for online access – in a cloud for example – a hugely important development has been the ability to put descriptions of the data – so-called ‘meta information’ Information about information) – in formats that machines can find automatically. This allows analysts, or data analytics systems, to determine relevant information in real-time, creating a virtual supply chain from many different sources that can be put together on demand.
This where data lakes come in. As virtual cloud data storage facilities, they are capable of holding vast amounts of big data. In theory, the billions of devices now connected to networks can feed their digital information into these sophisticated repositories for the purpose of processing and extracting insights. The same online retailer can store data transactional customer data with tweets in which customers have provided feedback on certain items, returns or offers.
Collectively, these four elements have produced a new world of data analytics that helps form the
basis of better decision-making – something that was simply impossible only a decade ago. But to truly compete with the world’s leading insights-driven organisations, business leaders must first develop a bird’s eye view of their data, where systems aren’t just integrated, but business functions also share in a common strategic goal. Whatever is coming over that hill, their survival may depend on it.
Ashok Suppiah is the founder and CEO of Mitra Innovation, a specialist in digital transformation and software development. www.mitrai.com
