This post is part of an ongoing series that explores how organizations can use data and analytics to drive performance outcomes.
A simple way to think about the different types of analytics an organization can use is to focus on three broad categories: descriptive, diagnostic and predictive. Each is distinct in the value it provides and in how it can be used by organizations to driver greater performance. The key is getting your organization to understand how they are different and when to use each one.
Descriptive analytics are within the domain of reporting and summarizing information. These analytics are focused on what is happening or has happened, for example: "Here are the top 15 salespeople in terms of revenue for the month of March in the Western region." Descriptive analytics can be very useful, especially for organizations that are just starting to wrap their arms around all of the different data being generated internally.
As part of their new analytics effort, one company we recently spoke with uncovered that they spent nearly $20 million annually on various leadership development programs. The end result was nothing more than a descriptive report of total dollars spent, but it had an impact because they learned how much they were spending on these programs. They still didn't know which programs provided the greatest ROI, but they could implement a plan to analyze which programs should be dropped and which should be replicated and expanded for the greatest return.
Diagnostic analytics are less focused on what happened and instead focus on why something happened. These analytics generally look to highlight processes and causes, rather than outcomes -- "Sales are down in the Western region and the likely reasons are a decline in prospect volume, attrition of our two top salespeople and a new entrant into the market." Descriptive reporting can't provide answers to the questions "How do we fix this?" or "How can we improve?" That's where diagnostic analytics come into play.
We see this clearly in Major League Baseball, where many teams have moved away from evaluating pitchers by how many runs they allow. Research has shown that the rate at which pitchers allow runs (outcome) is much less consistent, year-to-year, than their rate of walks and strikeouts (process). In fact, this year's runs allowed are less closely associated with last year's runs allowed than with last year's walks and strikeouts. By focusing on the process rather than the outcome, teams are better able to evaluate the talent they have and evaluate which pitchers to keep, trade and acquire. Considering it costs an average of $8 million per year to sign a free-agent pitcher, knowing whether a pitcher's performance truly was due to his talent and skill as opposed to randomness or luck can make a huge impact.
Finally, predictive analytics take the lessons learned about why things have occurred and builds models to predict what will happen in the future. These predictions can be in the form of macro-level forecasts, things like: "Sales in the Western region will grow by 5% to 8% over the next 12 months." Or the predictions can focus on the micro-level: "There is a 75% chance that our largest customer in the Western region will leave for another vendor in the next 12 months."
Forecasting can be extremely helpful as organizations look to make their operations more efficient by trimming costs. The process also simultaneously ensures organizations retain the resources necessary to take advantage of future opportunities.
More granular predictions, such as customer-level predictions of intent to repurchase, can be used to prioritize internal resources and inform support strategies from a sales and customer support perspective. For example, Gallup works with a number of clients in the franchising space whose growth is largely dependent on independent franchisees. We have worked with these clients to build models that predict which franchisees intend to build with the brand again. Armed with this information, the franchisors can better allocate their resources to the most valuable franchisees and those with the greatest potential for growth.
Unfortunately, studies show that most organizations rely predominately on descriptive analytics. A handful more use diagnostic analytics, but when it comes to predictive analytics, far fewer have made any real progress, outside of some rough forecasting.
Organizations that get the most out of their data intentionally move from one level of analytics to the next and understand the best way to use each in practice. This commitment to using analytics appropriately and strategically can help organizations gain a competitive advantage.