Analytics Value Now! — Opening Statement

Posted June 21, 2021 in Business Technology & Digital Transformation Strategies, Data Analytics & Digital Technologies Cutter Business Technology Journal
CBTJ Analytics
In this issue:


Over the last 20 or so years, many organizations have turned to analytics to deliver business value, drive digital transformations, and improve their competitive advantage. Most are not successful. A majority of analytics projects continue to fail to deliver the expected ROI. Reports on success rates of advanced analytics capabilities (including cognitive, machine learning [ML], and more) suggest even lower success rates.

As analytics leaders continue to expand their breadth and impact in organizations, emerging from the depths of the IT department to sit at the table with the executive leadership team as chief data officers and chief analytics officers, collectively we are charged with improving the value that analytics delivers to retain those seats. But with those leaders becoming more experienced from a technical and business perspective and our tools and technologies becoming more advanced, why do organizations continue to fall short of this goal?

It begins with the common mistake of misunder­standing the term “analytics” and what it involves. Some see numbers attached to any persuasive argument as the use of analytics when these values are basic mathematics (e.g., an average or percentage), basic business intelligence values (e.g., the number of customer orders last month), or both (e.g., percent of customer orders delivered on time). Others believe analytics is a silver bullet that can magically leverage data to solve any business problem using ML — though most don’t actually understand ML at all.

So let’s begin with a clear definition of analytics: the use of advanced mathematical techniques to identify, explain, or predict patterns in data that improve decision-making confidence to create value. Some analytics are simple and effective (e.g., regression, segmentation); some are extremely complicated (e.g., cognitive, ML, deep learning).

Measuring Value

What is most important in the understanding of analytics and its ability to drive organizational value is the identification of a problem where decision-making confidence must be improved. Albert Einstein has been quoted as saying “If I had an hour to solve a problem, I’d spend 55 minutes thinking about the problem and five minutes thinking about solutions.” With respect to analytics, he couldn’t be more accurate. Far too often, people jump to a solution because they want to leverage cognitive capabilities or develop a complex ML algor­ithm because it seems trendy or impressive — even when a basic analytic technique would solve the prob­lem. Or they’ve become enamored with the ideas of predictive or prescriptive solutions when a diagnostic or descriptive analytic solution better meets their needs.

Once this is clear, we can measure value more directly. For example, a retailer trying to identify the optimum price for a product that drives the greatest revenue can calculate the potential value of that analytical model as the difference between current and projected revenue. Attribution is always a complicated endeavor in retail, so the realized value of the model could be debated, but the potential value can be clearly calculated. Then, by factoring in the cost to build the model, we can see if the analytics created a value opportunity for the retailer.

Once organizations start mandating the calculation of potential value and the attribution of actual value to analytics projects, not only can we more clearly meas­ure the value, we can use those learnings to focus our efforts on future, higher-value opportunities. And by assigning accountability to business leaders for the realization of those benefits, organizations can increase the probability that the potential value will be realized. For lower-value opportunities, it is better to rely more on more subjective measures like experience or intuition — or employ basic mathematical techniques (e.g., extra­polation) to provide decision guidance.

Beyond this seemingly basic approach (one that too few organizations follow), there are myriad approaches that can help analytics projects deliver on their promises and potential.

In This Issue

The five articles in this issue of Cutter Business Technology Journal provide insights into some of these additional approaches.

The first article, by Avishan Bodjnoud, highlights the typical challenges of skill sets, data literacy, data governance, process, resources, and senior leadership buy-in that affect most traditional organizations. Without addressing these challenges, she explains why organizations that simply establish a standalone analytics entity and expect immediate results are often disappointed. Bodjnoud closes with an important discussion on the need to understand the reasons for resistance to change and then manage them — critical elements of any analytics project or digital transformation effort.

Next, Cutter Consortium Senior Consultant Frank Contrepois continues the discussion of human emotions and our connections to data. He shares how companies such as Google and Facebook manage to convert data into money and how other organizations can learn from and potentially replicate the approach (or at least the outcome). But this approach comes with some warnings: more data is not always better. It can lead to clutter in your data warehouses, causing longer processing times for algorithms and confusing dashboards and visualizations. Excess data also increases expenses, reducing the overall value delivered. Contrepois closes by supporting his earlier point: focus on a defined business problem and act by building an analytical model that improves decision-making confidence.

Cutter Consortium Senior Consultants Michael Papadopoulos and Philippe Monnot then take a deep dive into ML projects. They address the “very powerful tendency to anthropomorphize ML and AI, imbuing it with human characteristics.” As we increasingly describe them in human terms, we often fail to make a critical distinction in the way humans and machines interpret the world. The authors also explore the unique experiences that enable humans to effectively under­stand data signals. Finally, they articulate a number of common pitfalls that, if avoided, can greatly increase the chances of success of ML initiatives.

In the next article, Manjul Gupta, Carlos Parra, and the late Brian Mennecke look at how to achieve robust analytics. They begin by identifying effectiveness as the first hallmark, echoing the previous critical objective around finding an important business decision to be solved in order to deliver relevant value. They describe efficiency (i.e., balancing the amount of data, com­plexity of models, and investment in analytics to be commensurate with the business value) as the next critical element. The authors’ third criteria for robust analytics is around minimizing risk by monitoring threats and opportunities in both internal and external environments. Finally, they describe ethics as the fourth tenet of robust analytics: “using analytics for good, and for making the world a better place.”

In our final article, Ben Porter uses several case studies to show how organizations have made progress in amplifying the value of analytics by demonstrating three actions: recognizing how value is created, focusing on delivering that value, and understanding the changes that must be adopted to ensure long-term value. He describes the four fundamental requirements for successful analytics projects (sponsor, tools, team, and project/problem) and closes with the critical assertion that value creation from analytics requires teamwork between IT, business, and analytics professionals.

When reviewing each of these articles, it is important to recognize that deriving value from your next analytics initiative is not simply a matter of selecting the approach that best resonates. It is essential to understand each perspective and connect that to the specific situation at your organization. The context of your situation, the analytics maturity of your organization, the risk/value proposition, and your culture will make some of these techniques more effective than others. 

We hope you find this collection of perspectives insightful, impactful, and inspiring. As you contemplate how to ensure that your organization derives value from your next analytics initiative, it is imperative to develop your own path forward, determining what tactics you feel will work best in your environment.

At a minimum, you must clearly define a problem that has value, is worth solving, and has the potential to improve your decision-making confidence. Then, following implementation, it is critical to establish accountability and track the results. Only then will you be able to accurately measure the value of your analytics projects and achieve “analytics value now!” for your organization.

About The Author
Dave Cherry
Dave Cherry is Principal of Cherry Advisory, LLC. As a thought leader, executive strategist, and speaker, he helps clients in the customer experience (CX) industry (that's everyone with customers!) define a CX strategy, enabled by innovation and measured/informed by analytics that drives deep relationships and connections with customers. Mr. Cherry is a member of the International Institute of Analytics Expert Panel, serves on three advisory… Read More