Resist 'garbage in, gospel out' doctrine -- defend data quality
The 'garbage in, gospel out' approach to business analytics may be a valid approach for doing big data projects, but CIOs should resist it for business transformation.
The old data processing maxim of "garbage in, garbage out" was a sound reminder to IT and business leaders that you cannot generate high-quality information from reports built on unreliable data sources.
Today, we have a new catchphrase that, in my view, does not serve CIOs well: "Garbage in, gospel out." The slogan reflects the belief by management and IT that enough data (even poor data) will, nevertheless, produce great reports, analytics and decisions. The thinking is that a large enough amount of data will overcome shortcomings in its values.
While a "garbage in, gospel out" approach may be acceptable for big data projects, it is much less effective when used to support operational activity and internal decisions made around performance, customer interaction, financials, compliance, HR and so on.
For business transformation, there could hardly be a worse time for this "large volume" idea to gain traction. As important as good data has been in helping to make good business decisions, going forward, the need for good data will reach critical proportions as companies refocus from cost reduction and wrestle for market share. The ability to have access to the data needed to identify vision, set strategy and determine business and technology evolution is once again becoming a driving force for business success.
Hundreds of databases, overlapping data
Just at a time when data-driven decisions are requisite for success, almost every company has serious data-related constraints in making decisions and supporting company growth.
However, just at a time when data-driven decisions are requisite for success, almost every company has serious data-related constraints in making decisions and supporting company growth. To a large degree, this is a result of the "best of breed" application strategy that has resulted in hundreds of databases that have overlapping data in different configurations. This data is seldom in sync from application system database to application system database.
To add to this complicated web of databases, much of the data is at risk of being out of date. Even worse, no one really knows what data is correct, what is up to date and if the data has been corrupted. This has created a situation where making simple changes to applications is a complex undertaking with an unpredictable outcome. So, while we have data everywhere, we often have little understanding of what is right, what is redundant, what conflicts with other data sources and where we should go for accurate information.
In attempting to deal with this situation, many companies have resorted to a statistical approach and look at averages in large aggregations of data, relying on volume and probability to determine which information they should count on. However, in many cases, this approach results in untrustworthy information.
Fortune 500 impact
Permit me to cite two widely sourced statistics: From 1955 to 2016, just over 60 years, 88% of the Fortune 500 companies have gone out of business. Unfortunately, this attrition has accelerated. Based on this rate, by 2027, new firms will replace 75% of the companies that were in the Index in 2011.
The question is why? It doesn't make sense to claim that 75% of our largest companies will go out of business in 16 years from poor leadership. The same goes for a failure to keep up with technology. Some of these now-defunct companies, to be sure, bit the dust because of incompetent executives, and, no doubt, there were a few failures that can probably be chalked up to bad technology decisions. But these numbers are staggering.
Here's another argument for the ongoing decimation of once-powerful companies. As you look at the traits companies must possess to succeed today, two attributes stand out: The ability to change quickly with limited business disruption and the usability of the company's electronic data. Of course, these are not the only ways to win in a fast-paced digital era, but these two attributes are close to the top of the list.
Of these two issues, I would argue that data problems have the greatest ongoing impact on the company. "Garbage in, gospel out" is not a viable approach for operating the business: Bad data leads to bad decisions, internally and externally. It is that simple. The inability to use data to adequately support customers drives them away, as does presenting erroneous data to them. Bad data erodes confidence in the company. The same applies to supporting internal operations and employees. Together, these issues form a type of constraint that is difficult to overcome as a company tries to change and remain competitive.
Large, complex companies like those on the Fortune 500 are often slow to change and are, sometimes, seemingly incapable of change. This is, in part, because they are unable or unwilling to isolate and correct discrepancies in data or to remove clearly erroneous data.
Call to action for CIOs
In upcoming columns, we will explore how CIOs can help ensure their companies live another day to compete. CIOs can start the process by asking the following questions:
Are business decisions being made on questionable data?
Are business decisions generally good or could they be better?
Would you rate the overall general data in the company databases great, OK, needs help?
Can people count on the accuracy of the data in the systems and, thus, the reports?
Are you in control of your data and its quality?
Do you know how it flows and transforms?
Is management, as well as the company's internal and external data consumers, happy with the accuracy and availability of your data?
Is data in sync among the various applications systems and their databases? Is the data up to date?
What data, from what sources, is good? What can be counted on?
What is the past and potential impact of data issues?
If you've answered "no" to any of these questions, chances are your company will not be able to change quickly or to deal with customers in a flexible way. And efforts to do so will deliver mixed results unless its data issues are addressed. Presenting erroneous data to customers in the different interactions is really not acceptable. Dealing with customers on the phone or over email to correct errors in their data is also not going to be tolerated and will become a competitive differentiator for companies that get it right and do so quickly. Word of negative customer interactions and customer dissatisfaction gets around fast, courtesy of social platforms.
It may be that one of those viral rants about a bad customer interaction with your company proves to be the impetus to invest in correcting company data and keeping it accurate. But the reason for action is less important than the fact that the company is doing something. Although the course of action will vary by company, the definition of that course and the creation of an investment business case represent a good starting point. I also recommend that this correction become part of an integrated long term strategic business and IT direction.
Next in this series on data and business transformation
Next month, we will continue with our discussion on the importance of data quality in preparing the business to evolve, specifically: the role that data plays in operational improvement and large scale business transformation using technologies such as robotic process automation and business process management software.
My purpose in writing this series of articles is to raise awareness of this critical constraint to business transformation and to share ideas. I believe in creativity and I welcome any suggestions on how you may be approaching this issue. So, please share what you are doing -- many need ideas and help.
