The analytics continuum

By: Steve Holder, National Lead Analytics, SAS Canada

Data is the lifeblood of a business. The ability to understand what that data means in a business context drives differentiation, competitive advantage, and a fatter bottom line. And we’re generating and capturing more data than ever before, in structured and unstructured forms.

There are many tools to help us understand this data and put it into a context where it becomes actionable information. And though the terms business intelligence, data visualization, and analytics are often lumped together or used interchangeably, in reality, they’re three very different, though related, disciplines. They exist along what we’ll call an analytical continuum.


* Business intelligence (BI) is the older of the disciplines. In fact, the term was first coined in 1865 by Richard Millar Devens, describing the ability to collect information and react to it accordingly. BI as we understand was defined by Garner Group analyst Howard Dresner in 1989 as a term for technologies that provide fact-based decision support.

BI allows an enterprise to drill down into past performance—by region, by store, by product, by sales rep, by any discretely collected data point.

* Data visualization (DV) applies a visual and graphical interface to the data. This allows a user to see things in the data that might be impossible to discover by poring over BI reports, especially as data volumes grow. The ability to visualize, explore and pivot on data is the key difference between DV and BI.

* Analytics gives a user the ability to anticipate and predict relationships in the data, for example, how a 20 per cent decrease in promotional budget in a certain region would affect revenue or who we should target a discount to in order to raise customer satisfaction. Broadly speaking, there are three types of analytics. Descriptive Analytics, are the simplest and are used to provide straightforward information or answer “What has happened? Predictive Analytics, ups the game and uses mathematical techniques to model and forecast the data to understand the future or answer “What could happen?” Prescriptive Analytics, take it a step further and use statistical models and machine learning techniques to optimization the possible outcomes and answer: “What should we do?”

The short version: BI is simple reports; DV is visual exploration; analytics is optimizing the outcome. In a retail context, for example, BI can tell an enterprise how many units of a particular stock-keeping unit (SKU) were sold in July; DV can help determine that sales are affected by customer vicinity and promotional activity; analytics on the other hand predict that a specific promotion delivered to customers in a particular postal code will drive up sales while other promotions will not. Analytics provides the why behind the what.


If you use Google Trends as a barometer, interest in BI is waning; the use of the term has been in a slow but steady decline. By the same token, DV and analytics are on the uptick yet all three have relevance and applicability for organizations today.

Ask a handful of enterprises where they are on the BI-DV-analytics continuum, and you’ll get different answers, even within the same enterprise. Truth of the matter is, many organizations are using all three disciplines for different departments and different purposes within the enterprise. For example, logistics may be using simple BI tools to determine the frequency and number of deliveries to each store, while marketing is using an advanced analytics platform to predict the sales impact based on promotions used to boost sales. And that’s ok, but it’s important to create a closed loop decisions system so that the forecast funnels back into the BI solution to ensure logistics and manufacturing are prepared to deliver.

That’s one reason an advanced analytics platform has to provide support for all three disciplines in a fashion that is accessible to users with a variety of levels of sophistication. Another is the rise of a new population in the workforce.


In the last few years, several universities have developed data science programs. Generally a post-graduate degree, students tend to have an MBA or undergraduate degree in a statistics-heavy discipline. But they can’t churn out grads fast enough to meet the burgeoning demand for data scientists. This makes data scientists a rare—and expensive—commodity.

Enter the citizen data scientist.

The citizen data scientist is a frontline employee with an understanding of business problems, processes and a natural curiosity around data. He or she understands how different parameters within the business affect each other, and can spot patterns that can be leveraged to optimize business processes.

The citizen data scientist is helps organizations scale as they are concerned with analytics at an operational level. This frees up the more formal data scientists to focus on more complex problems, rather than day to day plans.  A citizen data scientist is not an IT guru, though he or she is likely a sophisticated or power user of technology who understands the data and the problem. And citizen data scientists are not statisticians; the details behind optimizing statistical modeling is beyond their purview.

This is a valuable population within the organization for driving efficiencies and competitive advantage. Enabling them requires a platform that abstract or automate the IT and statistical modeling elements from users, allowing them to focus on exploration and discovery and ultimately problem resolution. And since the citizen data scientist has to function at all three levels of the analytics continuum—BI (reporting), DV (discovery) and analytics (forecasting and optimization)—a platform that can seamlessly integrate all three is necessary.


For a deeper dive into the roles of BI, DV and analytics, watch our one-hour webinar, Data Visualization vs. Data Analytics. The webinar goes into greater detail on the characteristics of analytics, the barriers to adoption, how analytics can incorporate unstructured data—text, social media, etc.—and more.

And when you’ve identified potential citizen data scientists on your staff, SAS hosts quarterly hands-on seminars using our SAS Visual Analytics product. Contact Rebecca Gill at our Toronto office ( for more details.

As the National Practice Lead for Analytics, SAS Canada Steve Holder is responsible for driving the software go to market plan for SAS and providing customers with thought leadership around Analytics, BI and Big Data. Steve is focused on defining creative opportunities to apply SAS technology to drive tangible benefits for clients. @holdersm