Say goodbye to dumb data with these five steps
In an era of unfettered IOT, how exactly do businesses face up to the sheer scale of unstructured data? TransUnion Africa CEO Lee Naik explains.
You've probably heard the prediction that the world will have at least 50 billion connected devices as early as 2020. The actual number isn't quite as astronomical - realistically, we'll likely have between 20 billion and 30 billion, depending on who you ask - but that's still a mind-boggling number of opportunities for Internet of things (IOT) growth.
Hidden within those figures is something a little scarier - the sheer amount of mostly unstructured data that will be flying around. Cisco predicts we'll hit 2.3 zettabytes by 2020. That's an awful lot of information to make meaning of, never mind secure.
So, in an era of unfettered IOT, where every business needs to be a data-driven business to operate, how exactly do you face up to the sheer scale of the unstructured data before you? How do you stay afloat, never mind master it enough to see beyond the data?
Now that I've set the data-drenched stage, it's time to dive right in. While the future is uncertain and every organisation's data journey is going to look different, there are some universal pieces of advice I can offer to get it right.
Do ask the right questions
Start with the problem you're trying to solve, instead of setting off on some vaguely defined quest to 'get value from data'. It's important to get quite focused here, and ask specific questions related to your day-to-day operations. 'How do I stop stock loss?' or 'How do I prevent fraudulent transactions?', rather than 'How do I create business growth?'
Take Santam, which was able to save millions of rands in only a few months, just by asking how it could reduce the rate of fraudulent claims. Once you've got a problem, it's easy to correlate the processes and transactions that are involved. That way, you'll know both where you need to mine for information and what tools you can use to do so.
Don't build what you don't need
Should you, like Microsoft, go all out and build massive data centres to process all that data you'll need? Data centres come with an incredibly high cost of ownership, are notoriously resource-heavy, and are rarely worth the amount you invest in them. All of this is only set to get worse as data processing demands skyrocket. Instead of trying to emulate Microsoft or Amazon, rather make use of their existing platforms and scale as needed. Or consider other options like co-location.
Start with the problem you're trying to solve, instead of setting off on some vaguely defined quest to 'get value from data'.
The same goes for other aspects of data analytics. Building your own analytics infrastructure takes time and resources from your core capabilities, which makes no sense if there's an existing platform that does it better, and is more scalable to boot.
Don't go for the overkill
I know, I know. Data scientists are the wizards of the business world, with the power to predict the future and deliver magical strategies for success. Chief analytic officers are the hot newcomers to the C-suite, promising to transform your organisation into a lean, mean data-centric machine. But I'm going to go against the grain here and say your organisation likely has no business hiring high-level data science capabilities right now.
Not only are good data scientists in short supply and high demand, especially in South Africa, but most businesses are not in the right shape to provide the support they need to be effective. Unless you've already done the dirty business of capturing, cleaning and consolidating the right data, your hot new data scientist will struggle to find the right data points, never mind visions of the future. Stop over-focusing on data science and look at the architecture instead.
Do get your data ducks in a row
Alright, so now that we've put the data scientist on the back shelf, how do you go about making sure your business data is fit and fighting? Like any good fitness programme, it all starts with a hard look in the mirror and a health assessment.
Start with an audit to see where your data collection is at, and where it needs to be. Is there someone who can take ownership of data as a strategic asset, and drive investment and education? Do you already have the systems in place to capture and integrate your most essential customer data? Assuming you do, do you have the capabilities to measure, map and visualise it into something actionable? And is your executive aligned on what the outcomes of your data strategies need to be?
Don't do it all yourself
The reality is that most businesses don't have the depth of expertise or analytical tools to make sense of the avalanches of data they face every day. The good news is that it doesn't matter how you tackle the deluge - not to your customers or end-users anyway. They don't care what business intelligence software you use or who in your department knows how to work with Hadoop, only that you deliver on your promise to them.
Take advantage of public databases and platforms where you can - in the UK, many have started making use of the government's open source data portal, for example, to seek out relevant information to their industry. Wherever there are gaps in your analytical competencies, partner with someone who specialises in harnessing the power of data.
Remember, the journey of 50 billion connected devices begins with a single step. Ask the right questions, learn the right lessons, and you'll be on the right track to flying high in the era of IOT. What are you doing to get started?