How financial services can use data architecture to share knowledge across the business

Colin Gibson

Colin Gibson, Head of Data Architecture, Markets & International Banking at RBS, joins Process Excellence Network ahead of the Big Data Analytics for Financial Services Conference 2013, to discuss what big data and good data architecture means for the financial services industry. Colin also discusses the top challenges faced and explains the carrot and stick approach to data architecture.

PEX Network: What does the term big data mean for the financial services industry?

C Gibson: Talking from the viewpoint of banks, most banks have been dealing with some aspects of big data for many years. Big data gets talked about in terms of the Vs; volume, velocity, variety.

And in terms of volume we have volume of market data, price ticks, the quantity of risk results that we have to handle. In terms of velocity again, prices move fast and latency in data is very important to things like algorithmic trading. And most banks, most investment banks in particular, have quite distributed infrastructures and that causes variety in data and the challenges of combining data across multiple sources. Big data can also mean unstructured data, so web logs, documents, social media, etc, and the challenge of linking unstructured data back to structured data. So again we, for instance, have pockets of use of tools in all of those areas and have had for years.

And big data isn't a thing; it's a label that is currently being used to cover a wide set of tools that can be used to address specific business problems. And really the important thing for an organisation like ours is to really not to drive the solution from the tools, but drive it from the priority of the business problems and select the right tool for the job.

PEX Network: What does good data architecture mean for the financial services sector?

C Gibson: Obviously big data architecture is not a financial services specific thing. It's hard to get a good definition of data architecture; it's one of the things I talk about in the presentation at the forthcoming event. But it's like real architecture. Data architecture is a combination of the thing that you look at and the discipline that you use to arrive at the thing that you look at. So good data architecture is having an orderly arrangement of parts - the current and future-state blueprint of your data in the organisation - and applying the discipline to refer to that blueprint, follow it and keep it up-to-date.

PEX Network: Big data will mean a change for data architecture. Can you elaborate on the challenges and opportunities this provides in your perspective?

C Gibson: I'm not really sure big data does mean a change for data architecture. What I tend to say on this subject is it doesn't matter how big or how small or how normal your data is, if you're going to leverage it and exploit it and get the maximum value from it you need to understand it. And having good data architecture helps you understand your data and therefore is a foundation for big data to be successful. So I don’t think it actually changes it; it probably just is yet another thing that emphasises the importance of it.

PEX Network: And what do you consider the three biggest challenges in developing end-to-end data architecture in financial services?

C Gibson: Picking three from all of the challenges is always a challenge itself. One is legacy platforms. All banks probably have a lot of legacy platforms, particularly where organisations have merged and taken over other organisations - that will generally set back their plans for a simpler architecture. Legacy is one of the things that have to be coped with.

The number of people involved in changing the infrastructure, the applications of a bank, tends to be large. And getting all of them marching to the same tune and marching in the same direction for data architecture is another challenge.

And then probably even bigger than both of those is just the pressure of real life. Projects have a focus on delivery and when projects run into trouble that tends to be when they take their eye off doing good data architecture and just delivering and taking shortcuts. So making sure that short cuts, if they are taken, are taken consciously rather than accidentally or subconsciously is another challenge.

PEX Network: Thank you. And in your view how can financial services use their data architecture to share and distribute knowledge across the business and silos?

C Gibson: That's a good question. It's all about sharing and distributing knowledge because generally in an investment bank you have silos of applications and distributed architectures and the key challenge that that then poses is bringing all of the data back together again.

There are many functions that need to have a common view of the data across all of those silos. Data architecture is key to that in terms of giving guidance on the common keys that should be used to refer to different types of things in data; having common reference data on stuff like clients and products and equities and bonds, for instance. And most important of all, actually, having a common understanding of what the different data terms mean so that it takes out, eliminates, any chance of misinterpretation of different terms being used to mean the same things in different areas of the business.

PEX Network: And, finally, Colin, your presentation at the 2013 Big Data and Analytics for Financial Services conference covers the carrot and stick approach to data architecture. Can you elaborate a bit more on this?

C Gibson; Carrot and stick; really the way I view it is the world divides into two people. You have people who spend a lot of time thinking about data architecture and you have people who don't. And one of the issues is that actually we need more people to think about data architecture.

So actually what I do in the presentation is start off by explaining why it’s important and the benefits that it yields. And then I'll talk a bit about the traditional, governance-led approach, which is the stick, where you basically try to force people to follow the rules and you check up on them and you have control gates and quality gates and things like that. What I'll argue is that that's not enough; actually you haven't got enough people to influence all of the people that you need to influence if you're doing it as a stick approach. And what you need to do is offer some carrots, offer some incentives, making it easy for people to do the right thing. And I'll talk through a few examples of what we've done to help with that and I'll ask the audience to raise their hands and give me more examples, that we can all get and take forwards.