TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

The Importance of Data Maturity Modeling

Company performance is highly correlated to data maturity, so focus on making plans to get to the next level in your data maturity model.

• By William McKnight
• March 6, 2020

Technology continues to move at a rapid pace. Delaying adoption of technologies and methods that are redefining "competitive advantage" can sound the death knell to a company's prospects. A company must be aware of what the technology marketplace is making possible and, with discernment, engage the process of adoption.

For Further Reading: Why DMaaS Is the Future of Data Management How Mature Are Your Big Data Practices, Anyway? Take an Inventory of Your Data Assets

Data has proven to be a competitive differentiator. Top performers realize they need all their data; clean data; well-performing, catalogued, and secure data. They've grown their data science programs to utilize such data.

Company performance is highly correlated to data maturity. Metrics show a positive association between data maturity and financial performance. Company maturity is data maturity today so it's important to know where you stand.

Data maturity modeling is understanding where you stand vis-à-vis the possibilities across multiple dimensions. It is a cold, hard, quantifiable, objective look at data use in a company, department line-of-business, or whatever scope is modeled. Although the ratings are interesting and inform urgency, what's most interesting should be what the next level in each category looks like and making plans to get there.

Figure 1 represents example categories of data maturity modeling.

 

Maturity Level	Level 1	Level 2	Level 3	Level 4	Level 5
Data Organization	criteria	criteria	criteria	criteria	criteria
Data Architecture	criteria	criteria	criteria	criteria	criteria
Data Technology	criteria	criteria	criteria	criteria	criteria
Data Processes	criteria	criteria	criteria	criteria	criteria

Figure 1: Outline of a data maturity model.

A Data Access Example

One aspect of data maturity is how data is accessed. It falls in the Data Processes category.

Data finds its true value in an enterprise when it is put to use. For most data access, that means a human fetches the data for a unique use or utilizes a predefined data access method, such as a report. The old saw about business intelligence is that it gets "the right information to the right people at the right time." Data usage is the sizzle to data's steak. It is a final step in data's life cycle and it must be done well.

However, the accepted paradigm of effective data usage is changing. Once it was exclusively through reports built by IT from nightly batch-loaded data warehouses, which replicated one or small set of source systems. Those reports were deployed to the personal computers of "users" in what seems like, in hindsight, a very heavy-handed and resource-intensive process, and a good description of Level 2 (Level 1 being predominately departmental spreadsheets).

Just getting a report to a user's personal computer on a regular basis is nowhere near the pinnacle of data access achievement. Organizations that fail to recognize that are leaving tremendous value on the table if they are stuck at Level 2.

A majority of users have become more accustomed to a targeted presentation layer. Dashboards represent an advanced form of information delivery, a more mature approach to disseminating information. Acting on dashboards that already have a certain amount of knowledge worker intelligence built into them moves the organization to real-time competitiveness (and takes them to Level 3). However, dashboards can still be manual and limited.

Adding machine learning discoveries to self-service dashboards represent a modern, highly mature use of data (Level 4) while level 5, for this criteria, would be automated discovery and immediate installation into real-time processes.

The levels are highly summarized here. The most important point is about advancement.

Advancement

You can't skip levels in any category, nor can you advance in one category too far beyond the others. Maturity levels tend to move in harmony across the categories. You are either moving up a category maturity or down a category maturity.

Attaining and retaining momentum up the model is paramount for success. Keep in mind, the model will evolve over time as more companies embrace becoming data driven and put pressure on the model criteria, normalizing what had previously been considered high maturity. No plateaus are comfortable for long.

You will ascend the levels of the model through concerted efforts delivering business wins utilizing progressive elements of the model, and thereby increasing your data maturity. Budgets seldom arise specifically for improving maturity. The skilled enterprise technology leader realizes this and manages these sometimes-conflicting goals, delivering both business wins and improved data maturity.

For example, if there is a predictive maintenance project being planned, but the methods are assumed to be batch data collection of maintenance records and SQL coding to look at what parts are approaching replacement, you might improve Technology and Process maturity by placing sensors on the parts, getting real-time readouts, looking at correlating factors, product availability, and risk management, and making smart decisions using machine learning.

Or suppose you're already into machine learning and a new project comes up. All projects to date have not had access restrictions applied to the machine learning models, so for this project, you can add access restrictions to the models and make an ongoing practice, which would take that criteria (data access processes) up to Level 3.

Retrofitting maturity into an enterprise's mundane endeavors is as important as creating new projects to advance maturity. However, the skilled enterprise technology leader is also in prime position in an organization today to create the projects that utilize advanced levels of data maturity.