A Path to Data Literacy
Unlimited Data | BY JAMES KULICH | 5 MIN READ
Generative artificial intelligence (AI) has taken over as today’s top conversation topic. As I said in my last post, this is the time to observe, learn and experiment with all that generative AI has to offer. This is also the time to plan for the long game – laying the groundwork for producing the kind of value these powerful tools can be used to create.
In this first post of a three-part series, I’ll focus on a fundamental part of that journey: enhancing enterprise data literacy. There are many good resources for learning technical data skills. I will offer some thoughts on a framework that will prepare you to tap those resources at whatever level of depth you wish.
Why Enterprise Data Literacy?
The arrival of ChatGPT brought sophisticated data capabilities to any organization’s front lines. While this was a dramatic step, the tools for tapping enterprise data of all sorts have steadily been making their way to the edge for a long time.
In simple terms, enterprise data literacy is about building strong data chops across an organization. This means different things for people in different roles. Technical knowledge is certainly one component, but data literacy that leads to real impact requires combining technical abilities with organizational understanding, communication skills and human empathy.
As David Chrisinger and Lauren Brodsky point out in Because Data Can’t Speak for Itself, their recent book on telling persuasive public policy stories, we humans are all about stories. So, as the title of the book’s first section suggests, a primary goal of data literacy is learning to “Tell Stories with Data about People for People.”
A Data Literacy Framework
One maxim I use in my courses is “get the right data and get the data right”. Before you begin any technical data work, ask why. What problem are you using data to solve? Can it be framed better? Is there a good match between your data and your larger goals? Would an investment in additional data be worthwhile?
With initial answers to these questions in place, you can move to an exploratory data analysis. Alex Gutman and Jordan Goldmeier offer an excellent outline for this phase of the work in their book Becoming a Data Head.
As they write, argue with your data by discovering the data origin story, how the data were collected and by whom. Are the data representative or are they biased in some way? Are there missing values and do they matter? What about outliers? Do the outliers represent mistakes that need to be fixed or are they meaningful signals?
Today’s tools make it possible for everyone to readily engage in this level of data exploration.
A Few Tips
Paint a full picture with your data, but don’t overload your audience. A simple view into how the data are distributed can provide depth of understanding. I recently reviewed some survey data that contained strong opinions claiming to speak for everyone. Two or three basic graphs showing how different subgroups of those surveyed felt about the topics made it immediately clear that two very different and conflicting stories were being told. Look for clusters.
Don’t jump to conclusions. If you stare at your data long enough, you will find patterns. Knowing that two quantities travel together, as measured by correlation, does not mean that one causes the other. The relationship might be accidental or both quantities might be driven by hidden factors. When you see an apparent relationship, ask why.
Broaden your field of view to explore adjacent data. For example, might insights from local restaurant reviews inform predictions about other forms of spending such as rent?
Find ways to transform opaque numbers into digestible story lines. A simple statistic expressed in easy-to-understand terms can help a complex idea stick. My pedometer app tells me that I have accumulated 12,965 miles since 2018. It also tells me that this is equivalent to six trips down the Mississippi River. Now I know how far I have gone.
Be honest. Don’t make a claim with data that you don’t believe or can’t support. Using data to raise good questions is an excellent outcome. Don’t try to be perfect but do aim to be accurate. Always keep your audience in mind. Conveying information is good. Having impact is better.
From frontline professionals looking to convince customers in new ways to managers seeking guidance for better decisions, the opportunities to tap modern data abound. Dive in to see what is possible. And, if you are looking to develop deep data skills, Elmhurst’s Master’s program in Data Science and Analytics can help.