Business Analytics 101: Data Quality

It’s one thing to say your organization values its data. It’s another to actually take steps toward ensuring your data is in the best shape possible – free of errors and duplicates, accessible and ready for action whenever you need it.

Based on conversations I’ve had with CEOs about their data issues, many consider data to be the lifeblood of their analytics initiatives -- but few pay attention to the quality or condition of their data until something bad happens. And by “something bad,” I mean major problems that crop up when you least expect them – like when a large insurance provider mistakenly solicits someone who is covered under another family member’s policy – all because the insurance company doesn’t have a unified view of its customers. Or when, in 2008, poor data control and inconsistent monitoring cost the French finance service company Société Générale $7 billion in losses. Those problems aren’t just embarrassing. They put a business at risk.

That’s why data quality is so important. Problems like the ones I mentioned above are the reason every type of organization, in practically every industry, needs to keep data management at the forefront of its day-to-day business activities -- not just as an afterthought.
Data quality, defined
Let’s step back and talk about the concept of “data quality.” We’ll start with a few truths:

• Organizations that have a clear picture of their customers, products, employees, vendors and partners are more empowered to make informed decisions that lead to success.
• As the saying goes, “Garbage in, garbage out.” Organizations can’t get that clear picture if their data isn’t reliable.
• Reliable data is only possible if the organization has a way to systematically streamline, organize and maintain it.

In other words, data that’s disorganized (or sits forgotten in multiple, disparate silos) is basically useless.
Of course, we’re not just talking about small amounts of data. We’re talking about all the data your business currently has, as well as the data that’s constantly flowing in via data entry systems and from business partners. And when you deal with such large amounts of data, it’s bound to have inconsistencies: names that are misspelled, addresses that were entered incorrectly or duplicates that found their way into the system despite your best efforts to weed them out.

When disorganized or inconsistent data prevents you from having a clear, true picture of your organization, you can’t seize new opportunities or make important decisions that drive business forward. Bad data hinders success.

Data quality techniques are vast and varied. One of the most difficult aspects of data quality is getting your automated processes to react the way your people would in a given circumstance – recognizing the inconsistencies, the duplicates and the data anomalies. This is something that is easy for people, but more difficult for automated operations. This leads to an intersection of analytics and data quality processing, using statistical and computational algorithms to investigate, remedy and monitor data. Data quality relies heavily on pattern matching and data analysis – ultimately leading in having your automated procedures operate correctly without the constant need for people having to make the decisions and remediations.

An ongoing process, not a one-time project
On a basic level, data quality ensures that your information is correct, consistent and successfully integrated so your organization can operate as efficiently as possible. But building high-quality data is not just a one-time thing – it’s not something you deal with and never revisit. In fact, data quality is not a technique, it’s a habit. Organizations that do a good job of managing their data know they can’t be reactive when problems occur; they can’t fall into the trap of only addressing data quality issues when the water gets rough.

Successful organizations treat their data as a strategic asset from the beginning, creating a data quality or data governance framework to help manage their information. In the end, they win – because they understand that without good data, they can’t make good decisions. And when they can’t make good decisions, they can’t mitigate risk, optimize revenue or control costs; ultimately, they can’t reach the level of success they aim for.

Getting started, making it work
An organization may realize it needs to achieve better data quality, but how does it get there without spending too much time or money – or without completely overhauling its current system?

I frequently tell CEOs that they can get their data to work for them without a major investment. There’s no need to completely scrap the solution they’ve been using for years. Instead, they can take small, manageable steps to achieving better data quality.

For anyone who’s interested in exploring data government solutions, here are three suggestions for getting the ball rolling:

1. Start with self-reflection. What’s your data management style? On a gradient scale, are you undisciplined, reactive, proactive or governed with the respect to the way you manage your data? Understanding the data maturity of your organization is the first step toward determining how you’ll manage data from this point forward.
2. Recognize that a successful strategy encompasses people, process and technology. In other words, taking a holistic approach leads to better success. Your organization needs buy-in from not only IT, but executives and employees as well. They all need to adhere to a documented, repeatable process to ensure consistent data across the enterprise – and the right technology (matching technology, data integration technology collaboration tools, etc.) is what ultimately enables the new system to work.
3. Take small steps. Achieving true data quality is an evolutionary process; it can’t be done overnight. I can’t stress enough that even the most mature businesses need to make incremental goals before moving toward the next level of data governance.

Parting thoughts
In my book, The Data Asset: How Smart Companies Govern Their Data for Business Success, I begin with a quote from Aristotle: “The whole is more than the sum of its parts.” This is a philosophy organizations should live by when it comes to managing their data; good data will drive their success. Once you define a methodology that works for you, you’ll discover that you can make decisions faster, achieve transparency easier and eliminate problems that could stand in the way of reaching your goals.