What is value?

Welcome to part four of my series on data literacy, skills, and tools. Part three is here.

In part one I defined data literacy:

Data literacy is the knowledge, curiosity, and experience to create value from data

Taking the definition apart piece by piece 

People, process, and technology are necessary to create value from data. Knowledge is the ability to use tools and technology to uncover that value. Curiosity is asking, "What if…" or "What about…" within the scope and timeframe of the analysis. Experience is our or someone else's subject matter expertise to understand the analysis, the glue that is the context of the project.

So what is value and what's the relationship to data maturity?

Value means something has monetary benefit, is important or useful. (Imagine what a 1998 model Furby still in box would be worth!) In data literacy the definition of value can often be found in the roughhewn maturity model. This data literacy (or data-driven) maturity model generally is built around four phases, with one building upon the last:

Descriptive -> Diagnostic -> Predictive -> Prescriptive

Before we dive into each data-driven and/or data-literate organizational or personal maturity model phase, keep something very important in mind: Each of these phases is dependent upon data veracity. Data veracity is the quality and accuracy of the data on which people, process, and technology are applied to create value. More simply put: garbage in, garbage out. While we need clean and accurate data keep in mind that we need available, relevant data. The data-driven and literate recognize that they won't have all the data. Sometimes the analysis must pare down the data to what's relevant versus available and other times the analyst must work with limited data where what's necessary for a full solution isn't available or is inaccessible.

Descriptive

This first stage of (organizational and personal) data literacy tends to revolve around the development and presentation of charts and graphs that don't say much other than to define the world at a point in time. That's not a bad thing. For example, "How many customers did we lose last quarter?" is the question and the answer is what descriptive analytics gives us, such as "42". (There's an attempted humorous reference there, somewhere.) The benefit of descriptive analytics is that it's a start: we're addressing the question, "What is happening?"

Diagnostic 

When the descriptive charts and graphs are provided the follow up questions of the next stage start with, "Why did that happen?" Another question might be, "Did we lose more customers last quarter than the quarter before? Or two quarters ago?" Answers to these questions may be explained by competitive market forces (a competitor made a direct play to poach customers), internal issues like a sudden negative change in our customer service, external factors like supply chain issues, A/B (champion/challenger) testing such as email subject lines driving open rates, or even timing: maybe the holiday season impacts the customer base in a way we've not taken into account before. The courage to ask these questions is an aspect of the analyst's curiosity. Courage to ask and answer hard questions empathetically is key to successful data-driven changes. The diagnostic phase is most likely when analysts begin to leverage statistics to measure differences and test hypotheses.

Predictive

After looking back--what and why--the next step is to look forward. The predictive phase of data-driven and literate maturity begins by asking, "What is going to happen?" In our example the question is, "How many customers will we lose next quarter?" At this stage the role of the analyst and the data scientist can start overlapping. Predictive modeling becomes the method by which value is created from data when we stop looking back and begin looking forward. Whether it's building profiles of our customers to better target them, predict the number of website visitors in the next quarter based on the factors that influence those visits, or predict how many customers will churn in the next quarter the predictive phase of data-driven organizations is a major goal. Predictive capabilities can reduce costs, improve success across all phases of the marketing funnel and sales, increase lifetime value, and generally improve outcomes for cost.

Prescriptive

The final phase of the data-driven capability maturity model is action. If the organization asks, "How can we not lose so many customers?" the data-driven response is to determine what is necessary to make that happen. This is the ultimate opportunity to build strategy leading to action: not only do we understand what happened but we have ideas as to why and we have an idea as to the drivers or levers that we can control that impact what we'll do to solve the problem, reach the goal, or test our hypothesis in the market. This final phase may require partnerships across the organization between analysts, data scientists, and business units. These are the greatest opportunities to build trust through partnership so the data-driven cycle can be leveraged over and over.

I tend to think of data-driven value creation as answering three questions: What? So what? Now what? They map to the four phases of data-driven capability maturity but are easier to remember. No matter how you remember the steps or phases the work of becoming and remaining data-literate is difficult. Building trust through empathy, context, and success make that much easier.

Next up: Coding 

Parts one through four have been heavily focused on setting the stage for individuals: definitions, skills, expectations, value, and a capability maturity model. Next up: tools. What is coding? What are low-/no-code environments? What's the difference? The metaphor for part five is the instrument the musician chooses to create music.