For any business intelligence (BI) solution to provide accurate results, the integrity of the data is crucial. The data of any organization depends on the data governance framework that it has in place. The following are some of the common issues that arise in a data governance framework—and possible solutions for the same.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
1) Linkage issues in a data governance framework:. There is no link between two different files of a customer. The portfolios don’t talk to each other, and the decision-making is difficult. For example, in many banks, there is scarcely any way to connect the credit card portfolio to the loan portfolio. Sometimes there’s the chance that a loan will be given to a credit card defaulter. In such a scenario, taking decisions such as approving a loan becomes tricky.
I would like to illustrate with a personal example. I missed two payments for a credit card so the bank blocked my card. After two weeks, I got a pre-approved loan offer from the same bank and I got the loan. This happened not because the bank didn’t have a good BI unit but because of portfolio independence and the lack of an integrated view of the customer—clearly a problem with the data governance framework. This is a challenge most Indian banks face even today.
2) Incomplete data:. Data needs to be complete for good and effective analytics that aid the decision-making process. Unorganized data causes plenty of issues when it comes to preparing data for analytical use. This is a problem faced by most Indian companies due to their faulty data governance frameworks. In my experience as a consultant to many companies from the banking, insurance, retail and other sectors in the country, there were several shortcomings in the data used for analysis.
Allow me to illustrate my point once again. A retail apparel brand wanted an analysis of its data to be done. The company wanted to develop a marketing strategy by analyzing its historical data. This data had been collected through the company’s loyalty card program. The data was good, but it lacked several necessary variables such as ‘time a call is received’ and ‘offers used to make purchase.’ This created a situation where customer sensitivity to the offers could not be ascertained, and the company hardly knew the best time to call the customers. The situation would not have risen had a well organized data governance framework.
3) Redundancy of data: Another issue which a data governance framework may face is the collection of redundant information by different departments of the company. This not only wastes storage, it also creates problems for the integration of databases. Such redundant data must be cleaned, and there are methods to do it, even though that implies extra work for an analyst. The problem is more severe if the entries are different for a particular person for the same variable. Such scenarios occur often, and they require additional input from the business managers.
1) Create a data team & data map: A company needs to have a data team which is empowered enough to tell other departments what has to be collected. Companies may use different names for it. Usually it is a specialized team within the analytics team or BI unit. This team is responsible for building a good data governance framework. ICICI has successfully done this, plus they have smaller teams for other data needs from cross-selling to database management. They decide what to collect and work on the strategy to collect this data from the market.
2) Right data collection: Data entry is just a point in the data collection system. There are two ways of data entry:
(a) entry by the company representative, and
(b) entry by the customer
There is no common standard followed by companies. Data collection depends on the judgment of the head of the analytics team.
There are also two types of collection methods being followed. One group tries to collect everything possible and leaves more work for the analysts. The other team tries to collect less and tends to collect only the absolute necessary data. Both approaches have their benefits and shortcomings.
The first gives more room to the analysts to analyze the extra information and to come up with non-traditional business models. But when the forms are too long there is hardly any data collected. A hybrid data governance framework has therefore evolved now. This is something like the ones Facebook and other tech companies use—there are some fields which are compulsory while many are optional. These optional fields are filled in at a convenient time.
Again, the first costs the company more in the short term, but the data collection is mostly better than user-generated content. However, it is not always possible to keep representatives at all points. Complex forms are painful for users, and the challenge is to get the best quality data. It is therefore advisable to make very targeted and smaller forms. There are some companies which give users the option to fill the form over a long period of time. This has worked well for most Web companies such as Facebook and Google.
About the Author: Bhupendra Khanal is the Chief Executive Officer of InRev Systems. He is an Analyst by heart and calls himself a data person. Bhupendra is a passionate blogger and maintains blog named "Business
Analytics". He has also co-authored the book "Demystifying Twitter Marketing".
(As told to Sharon D'Souza)