„Most enterprise data are inaccurate and incomplete.“ (Dun & Bradstreet)
There is a great discrepancy between how good we think our data are, and how good that data actually are. A majority of surveyed companies have high or very high trust in their data, although poor data quality costs vast amounts of money due to unnecessary extra work, missed opportunities and poor decisions. Data quality goes hand in hand with master data, the concept of having a single reliable source for important corporate data. One common example where companies struggle with managing master data is customer information – as customers move, change names or telephone numbers, this data is often not updated by itself and therefore quickly becomes outdated. Not to mention that there may be different views on who is the customer: the buyer, the one who pays, or the holding company in case we are dealing with Business-to-Business sales? The problems with not having control over customer data range from mistakenly writing the wrong names when mailing sales offers or sending duplicate mails (mildly embarrassing) to missing out whole customer groups when making targeted offers or sending sensitive information to the wrong place (very embarrassing and potentially very costly).
Master data is often spread out on different systems and locations, especially in companies growing mainly by mergers and acquisitions and in decentralized structures. It is commonly quite out of date and frequently lack a way to track historical changes. Moreover, data input controls are often absent, and a rapid growth of data creation only magnifies the data quality issues. Summarized, poor master data harms any efforts to become a data-driven organization as trust in the data is easily eroded. According to a survey by Dun & Bradstreet, the following challenges are mentioned as the most important:
Mastering your data
There is however a way forward even though the task can seem overwhelming. As for so many other governance-related topics, it starts with ownership. Responsibilities and roles must be defined, and all master data should have an owner with mandate to decide about definitions and naming conventions. Integration efforts should be made to provide the master data as a technical service to the rest of the organization, in some cases also syncing it with other systems. Some data can be purchased, such as customer address details or geographical data (settling debates such as whether to write U.S., USA, U.S.A. or United States of America). One should also set measurable, concrete objectives for the initiative and follow up on the progress.
A master data initiative can never be a one-time effort. Even with the most perfect master data systems, it is a battle against time. Data quality work is a never-ending task. Frequently made, it is luckily much easier, as long as root caused issues are properly investigated and solved. It is also a domain with well-developed frameworks and methods, so guidance and best practices are available.
Having reliable master data – single source of truth – is a great help for successful Self-Service analytics. Trust in the data is of huge importance to spread analytical thinking and behavior across an organization. And when it comes down to the dollars-and-cents, well-governed master data vastly reduces the data cleaning and correction efforts needed by every single business analyst using it.
Want to know more about how to become the master of your data? Feel free to contact us – as digitalization experts, we may have some knowledge to share.