This is part 4 of a multi-part series on the Analytics Operating Model.

As we move forward in our blog series on the Analytics operating model, we pinpoint the essential processes for delivering analytics. Data may be a prerequisite for most of these processes, but a successful solution is born from both science and art. Unlocking business value requires a healthy dose of creativity to work with data in its native state – incomplete and confounding.

Pierre Teilhard de Chardin, a 19th century Jesuit priest and philosopher, may have been one of the first to encapsulate our need for creativity: “Our duty, as men and women, is to proceed as if limits to our ability did not exist. We are collaborators in creation.”

An elusive, but extremely important question in the advanced analytics journey is how to push past the limits and consistently discover new insights.  In one case, valuable insight may be revealed from a clustering algorithm which diagnoses data flow anomalies and points to a potential corporate network breach. In another case, a supervised machine learning model may be required to drive down false positive fraud alerts for consumer credit.

NA different path, guided by core analytics processes, will bring the kind of value that elicits praise from executives and shareholders.ew insights may come from retracing conventional approaches, data, or work processes, but it is more likely that a different path, guided by core analytics processes, will bring the kind of value that elicits praise from executives and shareholders.

The core processes that drive advanced analytics are shown in Table 1. We will focus in on the two processes that enable your analytics team to get the most from your data: use case and algorithms and modeling.

Table 1

Table1: Core analytics processes brin­g consistency to the new challenges of each new analytics project

Use Case

The use case defines the question or problem posed by business stakeholders: For example, “Can we reach our target market share if we adjust promotions for newly inactive accounts?” A use case is specific to an industry, a company, and even a function. While a problem may be widespread across an industry, the nuances of a company’s operations and organization require specificity. The use case may start as a single question, but is then iteratively refined to understand potential value, needed data sources, and viable solutions. It is both an initial scoping tool for the analytics team and a communication tool – Primary stakeholders must readily understand the use case if you are to gain their approval and support.  Stated differently and borrowing a phrase from Albert Einstein: “If you can’t explain it simply then you don’t understand it well enough”.

Although the use case serves as the initial direction for analytics, it must be validated periodically. The process of data discovery may reveal the need for different solutions. Think of the use case (what will we do?) as the first key to unlocking business value. The algorithms and modeling process (how will we do it?) is another key.

Algorithms & Modeling

With clearly-defined use cases comes the need to select an algorithm or other modeling technique. The results guide progress through data discovery, much like a navigation system assists drivers to an unfamiliar destination. We show some of the most common techniques in Figure 1.

Figure 1

Figure 1: With a wealth of algorithms available, understand key differences before selecting one


The algorithm best suited for a use case is driven by two considerations: (1) the type of data available and (2) the skill set of your team. In Table 2, we summarize the more common algorithms and call out data considerations.

Table 2

Table 2: By comparing common algorithms, you match data and need to best option(s)


The second consideration for selecting an algorithm is the experience of the analytics team. This “human factor of analytics” should not be taken lightly. If the team has limited exposure to different algorithms, their effort may be best focused on a simple descriptive or diagnostic solution. With the right training and support, a team can build up familiarity with more advanced algorithms. (Learn more about team roles in “Bring the Horsepower”, Part 3 in this blog series.)

Interpreting Results

Aligning a use case with an appropriate algorithm and models is both science and art, especially when it comes to interpreting the results. Scholar and Stanford professor, Donald Knuth, may have summed it up best when he said, “Science is what we understand well enough to explain to a computer; art is everything else”.  Advanced analytics lives at the intersection of business context, math & statistical know-how, and data-wrangling. This is where we uncover new insights, inform data-driven decisions, and ultimately push our creativity.

One of Pierre Teilhard de Chardin’s hopes was to play a role in developing the greater good for all of humanity by uncovering insights with science and technology.   We are in the nascent, but rapidly unfolding stages of using complex analytics to unlock value across most every industry. Join us as we build and navigate a roadmap for this journey: it’s sure to be a fascinating one.

Click here to learn more about our Analytics & Information Management practice. Want to continue the conversation? Contact us at, and follow this series as we dig deeper into each element of the Analytics Operating Model.