‘Data Management Should Precede Analytics’
Enterprises are facing challenges in coping with huge amounts of data, and in getting some serious insights from it. There is too much focus on analytics, without getting the fundamentals right. And BI is taking a backseat as business leaders want to know what will happen ahead.
BW CIO met Partha Iyengar, VP & Gartner Fellow to ask him about the challenges that organisations face, in BI/analytics. He talks about a three-pronged approach to get round these challenges. Partha also spoke about the role of the Chief Data Officer (Cdo) and the Cdo role in driving a data culture within the organisation.
Excerpts from the interview:
BW: What are the barriers to adoption of analytics that enterprises face today? What is the Gartner prescription (3 approaches/enablers) to get around this?
The general desire is to move away from the rear view, the BI view of the world in terms of using data to analyse what has already happened, to predict analytics – to looking at what is likely to happen in the future. There is a tremendous interest in doing that, but there are a number of barriers to taking that interest and converting it to reality.
Before you can do that, you need to have a very strong data management function. That is still something that is being put in place. Who owns the data? We have talked about information as an asset. Converting that into reality is something that enterprises are putting in place.
Enterprises that are further along in that journey in terms of structuring the data and putting together the data architecture and treating data as an asset -- is more likely to start addressing the needs of whether it is marketing, sales or different business leaders – all asking what is happening right now, or what is likely to happen. So the rear view of BI is less interesting these days.
Approach #1. In order to do real-time analytics based on the increasing volumes, velocity and variety of data you need a data infrastructure – Hadoop, HANA or some real-time database. That is the second barrier that enterprises face in terms of making those investments. We are seeing more enterprises making those investments. We see CIOs struggling with the fact that business decision makers are asking for analytics, but by the time they get them to those people with today’s technology the decision is already passed.
Approach #2. The infrastructure upgrade in order to provide real-time decision making in terms of meeting that need for actionable insights is the second step that has to fall into place.
Approach #3. The third barrier is the organizational structure and the governance, which has to come from the top. The data ownership is often in silos – HR and Finance own their own data. The cross organizational view of data, which is often where some of the interesting insights are to be found, putting in place the governance mechanism which allows that. It is about having a strong data management function that cuts across the enterprise.
With those three enablers, different organisations are at different stages of evolution. Some of the banks are quite matured in all three areas. HDFC Bank is probably the furthest in terms of using data and analytics for driving real-time decision making, and for conceptualising future products.
There are others that are more type-B in nature (mainstream companies who are close to the leading edge), and then you have type-C (laggards).
BW: How prepared are organizations in terms of the readiness of the data? I am talking about things like data validation, data cleansing, and data integrity.
That is a big challenge and it is really where the data management function comes in. The second thing is applying standards. Even if you cleanse the data once, five years from now it should not reach the same stage as it is now in. There are also data integrity and data cleansing issues. That kind of effort has to precede any serious efforts at analytics. The worst thing you can do is make decisions based on bad data.
Enterprises that are on that journey of the three steps/approaches that I mentioned earlier; the whole data cleansing is an intrinsic part of that journey.
BW: Why is having a Cdo (Chief Data Officer) so crucial to the organisation today? What would be the role of the Cdo? And what should organizations keep in mind when appointing a Cdo?
The three things that I talked about earlier – data management, data cleansing, cross-ownership of data – that’s the expected role of the Cdo. The Cdo is supposed to provide the cross-organizational view and create this vision of data as a corporate asset as opposed to data owned by silos (departments).
Secondly, he is involved in creating the data architecture. Enterprise architecture is at three levels: business architecture, information architecture and technology architecture. The DBA (Database Administrator) looks at the technology architecture. The Cdo should be looking at the middle layer, the Information architecture. Based on changes in the business what are the changes in the information needs of the enterprise, and then interact with the DBAs and the technology architects in terms of saying how do you translate the changes in the business architecture into changes in the information architecture, so that it drives changes in the technology architecture.
The next step after digital business, which will happen 3 – 5 years from now, is really moving into algorithmic business. There is a tremendous amount of interest in algorithmic business models and even algorithms as a source of revenue.
The Cdo has a key role to play in creating a portfolio of all the algorithms that exist in the enterprise. Every enterprise has many algorithms and they just don’t know that they have those algorithms. They have to manage the process of seeing which algorithms have value, to drive customer experience or even revenue. That is the evolving and future role of the Cdo.
