Data is crucial for success. But you don’t know what you should know if it’s not structured….
There is much evidence that data is growing fast and furiously. This is creating vast pools of information that is not being classified. Unstructured data can’t be mined, let alone understood. Is this a serious issue? If so, what can be done about it? We posed these questions at our roundtable discussion.
Is unstructured data a growing problem?
Yolanda Cornelius, technical team leader at Discovery: Anybody who doesn’t have unstructured data is very lucky. Up until two or three years ago, we were in the same boat where we had vast unstructured data, but we’re better off now. It takes changing the mindset of business to work towards classification and identifying what they are saving and where. But there is still a big learning curve.
Belinda Milwidsky, head of IT at Fluxmans Attorneys: Smaller companies are in a worse position, and are not necessarily getting the help they need from the IT industry. Enterprises have done a good job in their classifications strategy. The challenge is in the SME market, where we don’t necessarily have the budget to cope with this amount of unstructured data. The people who make the decisions don’t necessarily realise the risk. So how will SME markets deal with this amount of data?
Guy Taylor, head of data driven intelligence, Nedbank: The way it runs is on a use case basis. We all have unstructured data. The way I think about it is: how to identify the unstructured data you’re willing to keep and is it going to be an asset? Everything is unstructured until you structure it.
Nathan Desfontaines, Managing Director, CyberSec: It’s the rubbish, the reports and the spinoffs from systems and the log outputs and Excel spreadsheets. It’s all that. If I had a datastore that only had the right data in it, and I didn’t understand it, then I’d gladly go on the journey and get intelligence from it. But it’s 80% garbage, 20% good stuff.
What’s driving the problem of unclassified data?
Bryan Botha, CIO of retail & business banking at Standard Bank: The conversation from our side is being driven more by the cost/size of the data than it is about caring what the data is and what to do with it. The better part of me wants to say there is so much intelligence there, but the reality is we’re tackling this because we find ourselves buying storage more frequently than any other part of our IT business, and that has to stop, or at least slow down.
Olwethu Sinxoto, head of information security at Wesbank: The problem is really complex. It’s not just file shares. It’s email, which is one of the biggest sources of unstructured data and we’re trying to deal with it. So this is not an IT problem. It’s a business problem. It touches technology, people and processes. If you don’t have data stewards, someone on the ground who is enforcing certain practices, you’ll never achieve classification. IT and business are not separate. So it’s a collective effort.
Karin Höne, chief information security and risk officer, Barloworld Logistics: The drive is that we buy businesses that have to be profitable. There is no time for going through an exercise to bring everything elegantly on board. So we end up with a lot of duplication of systems and data. We may become one business, but I don’t have a view overall of what exactly is happening end to end on the processes. The other challenge is business doesn’t always understand what they bought from a data perspective – every business brings data with it – and they don’t want to invest in the time and energy to mine it. Yet they ask IT for more information and for it to be more intelligent. You cannot get to that end state if business is not prepared to invest in finding out what they have.
How can we get the business involved?
Dumisani Hlongwane, senior manager of IT & KM, Rand Water: Because business doesn’t know where to start, they just expect IT to sort it out. But when you start looking at it, you realise it’s not all mine. Somebody in the business owns this – and that’s where we start.
Dr Stanley Mpofu, CIO at Wits University: Everyone at Wits thinks they know everything. I’ve used creating awareness first as my strategy. It doesn’t matter if you start with first principles – academics will find a reason to disagree with you. So I focus on creating awareness first. I sit with different people and explain why we want to do things in a certain way.
Julian Ramiah, Group information & security officer at Liberty Group: We IT-born professionals inherited that thinking, whereby we own the problem as well. But now we have the business dudes around the table, and the first thing we want to deal with is the terminology we’re using is business lingo. We start with a problem statement and the roles and responsibilities. Now we find much more engagement from the business, because we could quickly show them, using some Hadoop platform, a single view of the customer across the portfolio. So don’t tick a compliance box. Rather show the business value, where you’ll get more buy-in and budget.
Abdul Baba, head of IT, Kwesé Holdings: We need to understand the drivers of the business: why do they need this information? We need to cater for the data, but we don’t understand where we can add value and use this data in the right way for quick decisions around profitability and growth. But that’s how business is seeing it. IT has a different perspective: we need to mine and store the data. But we need to also understand the drivers from the business and that’s where the conversation should happen.
How can we deal with the problem?
Ebrahim Samodien, CIO of Enterprise Functions at Barclays Africa: We’ve pulled data out of tech and launched data as separate in the business – call it the data office. We’ve done a lot of work building visualisation tools, the backend, working with delivery teams, trying to understand what we have and putting that into a strategy or how we consolidate multiple enterprise data warehouses. We have a strategy to move to a Hadoop platform, but that may take five or six years. Our biggest challenge is to fasttrack our evolution, because our business is going to transform and change.
Russell Clarke, head of Business Intelligence at iDigi-Tech Infrastructure and Operations, FNB: We have many systems that do duplication collection. You have to look at the overlap and build that one master set. It takes a lot of putting it all together, defining which is more true than the other. Once you have those master sets, those are the ones everybody guards. You have strict change control. Because it is a thing that gives shared value, everyone focuses inwards. The more you bring those on, the better it becomes.
Mornay Walters, chief information security officer at AngloGold Ashanti: I agree with this approach. That sounds like coding discipline, where you’ve got your branches that you merge, and you have your source of truth that you work on. Then you discard what you don’t use.
Original Post Brainstorm Magazine