Organisations rely on data. Whether manually scrutinising spreadsheets or finding new angles in analytics reports, businesses pore over the details to find more revenue, greater efficiencies, and future resilience. Digital systems provide the opportunities to use and understand data in much deeper, broader and faster ways. In theory, technology should help every company become a data-driven institution.
But in reality, these are much harder ambitions to realise. Data systems and outcomes are often a drag on company performance and budgets, says Gary Alleman, MD, Master Data Management.
"Data is the model of the business, and if we don't get the model of the business right, then the business doesn't function correctly. And there is a huge amount of hidden costs that comes from data integrity problems that we work around. Data integrity addresses those problems and gets things right the first time," he says.
Nomonde White-Ndlovu, Bidvest Bank's CIO, adds that not addressing the problem will make things worse. "We’re all from organisations that want to drive revenue and efficiencies. Data is an asset we’re not always using correctly. But I think all of us are more invested in ensuring we get it right. The richness of data can help you as an organisation to either grow and excel, or to make the wrong decisions, if you're not interrogating that data properly."
Sometimes businesses can't articulate what the problem statement is.
Pinkie Meyiwa, Development Bank of South Africa
Poor and unreliable data not only leads to bad decisions, but can also erode trust in that data. Yet nobody consciously wants to produce bad data, so why does it happen? Nishan Pillay, director, ASI Financial Services, points to two problems. First, data systems and processes are often created by people who understand the technology and infrastructure, but not necessarily the business outcome at the end. Second, the people capturing data are often not the people who will use that data on the other side.
"They don't always understand why they're inputting the data. What happens then is you're going down the rabbit hole of solving data integrity that gets more costly as you follow the value chain. But you will keep on creating problems unless you go right to the beginning of understanding that single employee. Maybe a field doesn't work for them. Can they provide feedback that it doesn't work? A lot of problems can be solved if we look at the start of the data value chain," he says.
Companies need to retain the knowledge around the context of data.
Ravi Narayan, Standard Bank
The people expecting outcomes from data can often also trip up attempts to ensure healthy data lineage, says Pinkie Meyiwa, head of Information Lifecycle Management, Development Bank of South Africa. "Sometimes businesses can't articulate what the problem statement is, so they will think that the solution is a tool or a system, not actually translating what their needs mean in terms of data. Hence you find that now and again, there will be a migration onto a new tool. But then the new tool doesn’t service the data or business needs. They then easily end up with multiple data systems, yet the issue was not the systems. The issue was that they could not articulate what it is that they wanted from an output perspective."
A complicated culture
Data is not a singular element or even moves unilaterally. It can morph and combine to offer different things to different parts of an organisation. This often results in gaps forming across the value chain because different groups don't link up around the purpose of data.
"You've got to look at the lifecycle of data," says Jacques Kleinloog, Revenue Assurance head, First National Bank. "The knowledge at the input side is very important. It’s understanding what the data means. But does that knowledge transfer to the end-user of the data? We work with a lot of complex data. There are the guys who receive the input data into the staging tables, moving into a BI environment or data warehouse environment. And then you've got product and finance teams, which come with different types of knowledge to extrapolate what information there is so they can on-sell to customers. You must connect the IT people who don't have the financial knowledge and the financial people who don't have the IT knowledge. That's the missing link in most organisations."
Data in its raw form can’t necessarily assist you.
Dr Nomonde Mabuya, 3sixty Health
It's also important to manage the knowledge around data and data systems that sit at different stages. For example, the data analyst who works on installing a new business system might leave and take that knowledge with them, and other people in the value chain can't necessarily close that gap. Yet we often think this will happen, says Ravi Narayan, Standard Bank's head, Data Integration and Interoperability. "Companies need to retain the knowledge around the context of data, and support that with technology to create a channel from the originator across the value chain, and transmit that knowledge to every consumer."
The data value chain
Data lineage and trust are reinforced when there’s a healthy data value chain. Silos, fracturing and competing data systems remain the fundamental barriers stopping companies from creating and maintaining that chain.
As Prenesh Padayachee, chief digital officer, SEACOM, notes, it comes down to having a single version of the truth. “Typically, every department that needs data tends to recreate it in a different structure that they need. I think that happens because there isn't the right infrastructure for data to be stored in a fashion that allows people to slice and dice it."
Solving data lineage and trust issues extends back to the basics – not only at the start of the data value chain, but considering the organisation's data maturity, says Dr Nomonde Mabuya, chief operations officer at 3sixty Health. "One of the things I checked is what is the maturity of data analytics in the organisation? That's where I started understanding where these issues are sitting. What became important is that when we started that journey, we started with the data itself. What quality of data did we have to support our analytical strategy? That became important because data in its raw form doesn’t necessarily assist you. So we needed to define what quality of data was available to us to help us create the outcomes we aimed for."
Mabuya also emphasises the importance of looking at company dynamics, such as silos (are different groups talking with each other towards the same goals?), and creating data teams with champions but also wide expertise and room for new people to keep expanding knowledge and diverse inputs.
What does it take to fix data lineage and trust? Technology is important, but as Jonathan Robson, account director, Precisely, says: "Technology isn’t enough. It's about the culture across the organisation, having the business understand that they’re the custodians and the owners of data, and putting the infrastructure and service in place to support different parts of the business from input to output across the data value chain."
* This feature was first published in the October edition of ITWeb's Brainstorm magazine.
* Article first published on brainstorm.itweb.co.za
Share