The big data phenomenon has driven many organizations not only to increase analytics initiatives, but also to focus on improving data quality in order to make reliable decisions. After all, what good is a lot of bad data? Aberdeen says that when data is "unhealthy," it's "one of the root causes of inefficient, time-consuming business processes [and] inaccurate decisions."[i]
So what exactly have companies been doing to manage their data and improve data quality? Some have implemented enterprise-scale master data management (MDM) systems to centralize information and others have implemented extract, transform, and load (ETL) tools to profile and cleanse their data. The size of your company (and your IT budget) may dictate the options you have for managing your data, but there's always a way to ensure data quality, one way or another. Let's evaluate some of the options out there. Today we'll tackle MDM.
As the name suggests, master data management is a huge initiative. Its goal is to create a single, unified view of an organization by integrating data from a number of sources – a centralized master file that contains a single version of the truth. According to Aberdeen’s 2012 report on The State of Master Data Management, most organizations aren't doing anything crazy when it comes to MDM. In fact, 45% of companies (mostly SMBs under $500M in revenue) are simply using spreadsheets or individual databases to manage and store their master data. Others (21%) are using internal talent to develop home-grown MDM systems.
However, MDM is not just a technology-focused effort. There's a lot of internal discussion and consensus that goes along with MDM. Here's a good example – defining customers. It's amazing how difficult it is for companies to gather an accurate list of customers and much of that difficulty arises from a) differing definitions of "customer" across various departments and b) differing data in various data sources. At arcplan, we used to encounter this issue. The finance team had one system that marked companies as "no longer a customer" once they cancelled maintenance (thankfully this is a rare occurrence; we have one of the most loyal customer bases in the BI industry). However, the marketing/sales CRM left the same companies marked as "customer" because they were still using arcplan (just no longer paying maintenance). Once we merged the two systems and agreed on a master definition of "customer," there was no longer a discrepancy.
MDM can be a long, drawn-out process, but the benefits of data centralization are significant. They include streamlined operations and improved ability to share and analyze information quickly, which in turn supports better decision making – a valuable outcome for executives who want to act wisely and improve company performance. Aberdeen also reports that companies with an MDM system in place report a 9% increase in customer retention (compared to 3% for companies without MDM) and a 7% increase in overall sales team quota achievement (compared to 3% for companies without MDM). So there are tangible benefits to investing in MDM as well. See the chart above for additional performance gains.
With MDM, you can expect a few challenges. Sure, it'll cost you money but it'll also cost you time and effort. MDM system implementation requires a lot of data analysis, consensus-building, and policy development at the onset of the initiative. Nevertheless, the business value of an MDM system is undeniable and continually proves to be an effective way of managing data and improving data quality.
Next time, we'll take a look at how ETL works when it comes to resolving your poor data quality issues.
[i] The State of Master Data Management, 2012. Aberdeen Group.