Better Data Management Drives Better Analytics Results
Business success is measured by how fast you can deliver the next business insight that accelerates your organization. That’s true regardless of whether your goals are improved customer churn, better recommendations for sales optimization, or better patient health outcomes.
And success depends on the state of the data your analysts have to work with.
It’s often noted that information doubles every two years. So your analysts aren’t short of data to analyze; they’re short of data that is clean, connected, consistent, and timely. So much data is coming from outside of the organization and has little or no data structure, data quality, or business context (metadata). And that’s where you step in.
Taming the growth of data for accurate analytics
If your colleagues ever wonder why you do what you do at work, I’d recommend sending them a link to our latest SlideShare presentation. Titled “Becoming Data Ready in the Digital Age,” it does a great job of framing the overwhelming data management challenges presented by the explosion of data volume and complexity in our information age—as well as providing strategic responses. The visually appealing presentation explains the need to have a single, reliable data management infrastructure that fuels all your data projects.
As I click through the presentation, I find it has a lot of bearing on analytics in particular.
Your analytics are only as good as your data, and as organizations add big data capabilities to augment their existing—and growing—investments in traditional data warehouses and business intelligence (DW/BI), they run the risk of actually slowing down data delivery if they use different data management tools for structured data and unstructured (big) data.
Instead, organizations should plan for a robust, single data management platform that works with any data, and any type of application, or analytics use case.
Drinking from the fire hose
In many organizations today, data runs on two tracks.
In the DW/BI track, data is carefully modeled to give it structure, run through data quality processes, and is often given business context by a formal or informal data governance process. This results in high-quality data that is good for management decisions and critical business processes.
In the other, big data track, the emphasis is on innovation. Data is ingested into a Hadoop cluster or data lake with little or no preparation at all. The job of the data scientist is to look for useful business insights in these data lakes. The approach is to fail fast, iterate, and find new insights. The problem is that they must spend 50 to 80 percent of their time on data preparation, leaving very little time for the job of analysis.
Two tracks, with two different approaches. But when your data scientists hit on that one-in-10 experimental success, you’ll need to operationalize and enhance the data preparation steps so that you can produce these analyses on a repeatable basis, with data quality levels that can support significant business decisions.
Preparing the handoff
The answer is to design an analytics data management architecture that works across structured data, unstructured data, on-premise analytics and cloud analytics, all with the same tools and skills. Then, it will be relatively easy to operationalize the insights found in the data lake world. Often, this means moving the data and processes to the more mature DW/BI world. At least for that present, as the DW capabilities of the data lake mature.
There’s a lot more in the SlideShare presentation—it’s a quick deck, but it covers a lot of ground (often in threes):
- It looks at analytics and three other key challenges to data success.
- It also calls out three traditional, and still-dominant, approaches to data that will actually slow down your data delivery—just when the business is requiring faster data delivery in order to be competitive.
- And it closes with three simple principles that cover what your data needs, what your business users need, and what IT needs.
Check it out and then tell me what you think in the comments below.
View the SlideShare to learn about better data management for the information age.
Better data for better analytics enables businesses to compete better or even disrupt their markets. Web-based Tinkoff Bank is a great example of businesses that are applying competitive pressure for others to modernize their data infrastructure, to extract and act on insights more quickly. I’ll explain just how Tinkoff is doing that in my next blog.