Usamos cookies y otras tecnologias similares (Cookies) para mejorar su experiencia y proporcionarle contenido y anuncios relevantes para usted. Al utilizar nuestro sitio web, acepta el uso de Cookies. Puede cambiar su configuracion en cualquier momento. Politica de cookies.


Gaining Insight through Analytics – It Takes More than OLAP

Originally published mayo 6, 2008

I have said many times that business analytics must satisfy three criteria to deliver real value: they must be purposeful, insightful and actionable. This article focuses on the qualities that make analytics insightful. Business analytics enable insight when they lead to discovery of new facts and awareness of things previously hidden. I believe that most of today’s business intelligence systems fail the insight test – a failure of both technology and application.

Some Insight about Insight

To create insightful analytics, we must first understand the nature of insight. Insight is a clear and deep perception of a complex situation or condition – the ability to “see inside” the situation. Insightful analytics are those that create the ability to look inside deeply enough to truly understand circumstances and to have a real grasp of cause and effect. Depth of understanding encompasses the ability to see:

  • What has happened – and what to expect in the future.

  • When past events occurred – and when to expect future occurrences.

  • Where current conditions exist – and where future impacts are expected.

  • Why current conditions exist – and why future impacts are expected.

Considering insight from this perspective, it becomes clear that the mainstream of business analytics – OLAP, scorecards and dashboards – lacks the capacity to produce insight. These technologies may create desire and some limited opportunity for insight. Uncovering real insight in data, however, is a more complex thing. It depends on a combination of advanced analytic technologies and skilled people. The technology aspect includes such things as predictive modeling and data mining. The human component involves business analysts who have what I describe as analytic vision – the ability to see the story behind the numbers.

Analytic vision varies widely among individuals and is influenced by characteristics such as curiosity, intuition and persistence. Given a set of statistics and metrics, different people will see them in entirely different ways. Those with the most limited vision see only what they expect to see – a number, a trend or comparison with a target. A typical viewer with average vision will formulate a new question – knowing what they want to see next. The instinctively curious and naturally intuitive find the places where they want to dig deeper – areas for further investigation. These differences in analytic vision significantly affect the opportunity to gain insight as illustrated in Figure 1.

alt

Figure 1

Insight and Technology

Accepting the premise that insight depends on analysts with the ability to see beyond the numbers, then it follows that the role of technology is to maximize opportunities to gain insight. Ideally analytic technology should improve the analyst’s ability to see the story behind the numbers – their analytic vision – by acting something like eyeglasses for the myopic analyst.

Here lies the reason that most business intelligence systems fail the insight test for valuable analytics. The mainstream analytic technologies – OLAP, scorecards and dashboards – simply don’t have the ability to radically enhance analytic vision. As Figure 2 illustrates, these technologies don’t climb very high up the curve of insight opportunity.

alt

Figure 2

Dashboards describe what is happening now. But they know nothing of where or why, and don’t offer even a glimpse into the future. Dashboards effectively support seeing what we expect to see. They may occasionally lead to some limited ideas about next questions. In reality, dashboards are more a tool for executives than for analysts.

Scorecards have a similar focus on what, with limited view of where and why. With trend lines, they may offer hints about the future. Scorecards can help to determine the next questions – knowing what you need to see – but rarely lead to fruitful areas for deeper investigation. Scorecards inform management but offer little true analytic capability.

OLAP edges a bit closer to opportunity for real insight. OLAP is focused on what, is capable of where with a strong spatial dimension, but is still anchored in the past. OLAP is a true analytic tool that can help to determine next questions and to determine what you need to know next. It brings some ability to find new areas of investigation, but the capability is inhibited by data volumes and labor intensity.

Beyond Mainstream Technology

Don’t misunderstand. It is not my intent to say that dashboards, scorecards and OLAP are without value. Each has a purpose – dashboards to provide a concise view of the current state, scorecards to track performance against indicators and OLAP for multidimensional data analysis – and each fulfills its intended purpose. None, however, has the explicitly stated purpose of creating opportunity to gain insight – of improving analytic vision. For that purpose, we must look beyond the mainstream of analytic technology.

I classify insight-enabling technologies into two categories: (1) the “stretch” technologies that expand the range of what we can learn from an existing pool of analytic data, and (2) the “expand” technologies that increase the analytic data pool. Stretch technologies include predictive modeling, exploratory data mining and discovery automation. Expand technologies include text and spatial analytics. The effect of these technologies in climbing the insight opportunity curve is illustrated in Figure 3.

alt 

Figure 3

Predictive modeling uses data mining and statistical techniques to examine patterns in data, to infer causal relationships, and to predict future behaviors. Predictive models go beyond what has occurred in the past to seek why it occurred and to quantify probability of future occurrences. Credit scoring in financial services is perhaps the most widely known example of predictive analytics, but it also has application in telecommunications, retail, healthcare and many other industries.

Exploratory data mining seeks to uncover hidden patterns, relationships and anomalies by sifting through large volumes of data to discover that which is not apparent and obvious. These discoveries are a big step up the insight opportunity curve. They expose areas for further investigation and begin the process of learning what was previously unknown. Exploratory mining is valuable but can be a labor-intensive process that involves data sampling, exploration, pattern definition and validation/verification of patterns.

Discovery automation reduces the time and labor associated with exploratory data analysis using prepackaged data mining algorithms in conjunction with intelligent software agents to find and present interesting patterns, correlations and anomalies in a set of data. When combined with the concept of “guided OLAP,” discovery automation is particularly powerful for insight opportunity. The discovery tool presents interesting patterns and anomalies as a list of areas for further investigation. The list of patterns and anomalies provides direct navigation into an OLAP tool to support the act of investigation.

Text and spatial analytics increase opportunity for insight by expanding the pool of data that is useful for analysis. An expanded data pool certainly means new possibilities to find patterns and correlations, to find new areas of investigation and to discover previously unknown facts. Spatial analytics, in particular geo-coding of data, increases analytic capabilities in OLAP by adding breadth and depth to the location dimension – a real opportunity to get beyond what and understand where things happen. Radio frequency identification (RFID) brings yet another spatial opportunity to analyze data about the movement of things – the essence of location intelligence.

Insight and the Business Analyst

However powerful the technologies – mainstream, stretch and expand – they have limited ability to create insight and no ability to act upon that insight. The insightful business analyst has the ability to distinguish between correlations, coincidences and causes and to find clear and deep understanding of complex situations. They seek to get beyond what and to fully understand the where, the when, and the why of every situation. They seek connections between the past and the future without assuming that the future will simply repeat what happened the past.

Business analysts are the connection between data and insight. They use technology, but they need much more than technology. Insightful analysis is driven by curiosity, guided by intuition, and fulfilled through persistence and tenacity. These are the keys to real business insight. It takes much more than OLAP.

Footnotes:

  1. Discovery automation is a term that I use to describe a class of technologies that lack a popular name. Examples of discovery automation technologies include PolyVista and DataMiner.

SOURCE: Gaining Insight through Analytics – It Takes More than OLAP

  • Dave WellsDave Wells

    Dave is actively involved in information management, business management, and the intersection of the two. He provides strategic consulting, mentoring, and guidance for business intelligence, performance management, and business analytics programs.


Related Stories


 

Comments

Want to post a comment? Login or become a member today!

Be the first to comment!