Information Management Evolution: Endings and Beginnings

As I sit down to write this blog I have the urge to sound off about two things that I recently read. One is a Datadoodle post by Ted Cuzzillo titled Coming: The end of data literacy as we know it. The second is a LinkedIn post by Bill Schmarzo proposing that Data Mesh “is nothing more than Business Process-Centric Data Marts federated with Conformed Dimensions.” On the surface, these two posts appear to be entirely unrelated. But what I see here—maybe because I’m in a reflective frame of mind at the end of 2021—is two people whose ideas I respect looking at the field of information management from the perspective of beginnings and endings. 

I could wax philosophical (and bore the hell out of you) about endings and beginnings. Instead, I’ll offer some thoughts from Marc Chernoff to set the tone for what follows. Chernoff says “In every end, there is also a beginning. The secret to a good life is to pay more attention to the beginnings than the endings.” In that spirit, I’ll respond first to Ted’s post and then to Bill’s. 

The end of data literacy as we know it

Ted Cuzzillo and I have spoken often and at length about data literacy. He knows that I actively promote data literacy as an individual goal as well as a critical component of the corporate culture. I wrote the e-book Building a Data Literacy Program, and I’m an instructor and Director of the Advisory Board for eLearningCurve’s Data Literacy Certification (DLC) program. This work reflects my belief that we are in the early stages of the data literacy journey. We are at the beginning, and the end is nowhere in sight. 

Ted’s article arises from Tableau’s announced acquisition of Narrative Science. There is good reason to find this announcement exciting. Automated data storytelling–using technology to create a narrative from data– is a valuable addition to the world of data analytics, and an important recognition that storytelling is much more than just data visualization. But this is a beginning, not an ending—the beginning of a new way to turn data into information that does not in any way portend the end of data literacy. I would expect quite the opposite: automated storytelling elevates the importance of data literacy.

The fact that 67% of employees are not comfortable with data analytics
makes a compelling case to invest in data literacy.

Consider one quote from Ted’s article: “only 33% of employees are comfortable using data analytics to support their decisions.” It is reasonable to assume that those most comfortable with data analytics are the most data literate among employees. The least comfortable are those who don’t know where the data comes from, don’t know what it means, and don’t understand how inferences and conclusions are developed using data. Changing the presentation of analysis results from visual to narrative will not change comfort levels for anyone. Narrative doesn’t resolve the uncertainty about data sources, data meaning, and analysis processes. That’s a job for data literacy. The fact that 67% of employees are not comfortable with data analytics makes a compelling case to invest in data literacy.

The end of data literacy isn’t visible even in the most distant projections. Automated storytelling adds to the ways that we turn data into consumable information. It doesn’t replace reports, charts, graphs, scorecards, dashboards, etc. We still need data literacy to understand, share, and communicate with those kinds of information presentations. Now we’ll also need it for a new beginning in information delivery—automated data storytelling.

Don’t read these comments as a complete denial of Ted’s premise. There is at least a grain, and perhaps a bushel, of truth in his message. He doesn’t simply say “the end of data literacy.” He says “the end of data literacy as we know it.” That end may be a good thing because today we know data literacy as a poorly defined and largely unfulfilled need. If this end is also the beginning of data literacy as a well-defined set of capabilities integrated into organizational culture, then I fully embrace Ted’s vision of the future. 

Data Mesh is nothing more than Business Process-Centric Data Marts 

To his credit, Bill Schmarzo presents his proposition about Data Mesh as a question: “Is a #DataMesh nothing more than Business Process-centric Data Marts federated with Conformed Dimensions using the Enterprise Data Warehouse Business Architecture and empowered with Master Data Management?” That’s a good question that should cause us to pause and think. Bill goes on to say that it sounds like something Margy Ross and Ralph Kimball have been doing for more than 20 years. 

Bill makes a good point that brings back the concepts of endings and beginnings. The data warehousing of more than 20 years ago has not ended despite the many declarations that the data warehouse is dead (my thoughts from 2017 here). The Kimball approach to data warehousing is in many ways the beginning of data mesh thinking—independent data domains instead of rigorous enterprise-level integration. It is reasonable to see in Kimball-style data warehousing the beginning of data mesh—but only the beginning, not the ending. 

I disagree with the “nothing more” part of Bill’s proposition. Data mesh is much more than Business Process-centric data marts. We have evolved from that influential beginning to today’s Data Mesh architecture that has many differences from Kimball’s bus architecture:

  • Data mesh works with data domains, and not all data domains are “business process-centric.”

  • Data mesh shares data through data products, and not all products are dimensional data marts. 

  • Data mesh imposes interoperability standards that go well beyond conformed dimensions.

  • Data mesh integrates data governance into the architecture.

  • Data mesh includes shared data infrastructure including data lake storage, data pipeline management, and data cataloging — all things that have emerged since the peak of data warehousing. 

Data mesh doesn’t replace data warehousing, data hubs, data fabric, or other architectural frameworks. It is additive. 

In the spirit of Marc Chernoff’s thoughts, let’s think of Data Mesh as a new beginning in data management architecture and consider the changes that it may bring about. Data Mesh doesn’t replace data warehousing, data hubs, data fabric, or other architectural frameworks. It is additive and creates new opportunities to address the complexities of data management. (See Data Architecture: Complex vs. Complicated.)

Final Thoughts  

In closing, I want to say thank you to Ted and Bill for their thought-provoking posts. While I don’t fully agree with either of your ideas, you each made me think and respond. That’s how we continue to grow and evolve the field of information management—through ideas, thoughts, and communication. To everyone who reads this, please join in the discussion. Add comments here, or comment about Ted’s ideas at Datadoodle, or respond to Bill’s proposal on LinkedIn.

Dave Wells

Dave Wells is an advisory consultant, educator, and industry analyst dedicated to building meaningful connections throughout the path from data to business value. He works at the intersection of information...

More About Dave Wells