Business intelligence is the transformation of data into insightful, meaningful information that is useful to a business for making decisions and driving action.
At its core are four pillars:
- Technology platforms
It is critical that all four pillars are tackled in tandem when defining a business intelligence strategy. In my experience, I have seen clients begin with a focus solely on current state reports and/or the data warehouse. They neglect to understand how and to whom the information will be presented, or how the data may evolve, or the platform may scale. This leads to delays in getting the data out to the final consumer and ultimately a loss of faith in the system.
In the business intelligence world, in its simplest form, data is made up of measures and attributes.
Measures are quantitative and mostly numeric. Common measures are (i) monetary – like revenue or cost; (ii) quantity – like units sold, units produced, hours worked; (iii) calculated measures can then be derived, like profit, percent growth, production rate, utilization, etc.
Attributes are qualitative, and can be thought of as descriptive elements. One key attribute is time. Business users look at changes over repeating intervals of time and examine trends, rates of change, variance, cycles, exceptions, and outliers. Other common attributes are customer, product, and geography.
The myriad permutations of looking at measures with respect to their attributes can provide insights on profitability, attribution, cost, productivity and risk, to name a few.
Data can be backward looking, where it is used to examine past performance, or forward looking, where it can be used to build models and drive forecasts.
Organizations then set up business goals and key performance indicators (KPIs), which become the second layer of a business intelligence system.
The value of data is truly realized when it can be presented to business users in a meaningful way. Traditionally, these were pages and pages of reports and spreadsheets. I have seen a hundred-page report placed on a manager’s desk each day. With the glut of information available, today’s business users do not have the time or the inclination to sift through reports. Users have become more sophisticated and demanding, and wish to be able to visualize data so that it can be meaningful quickly.
To do so, they need the ability to see high-level information on dashboards, and then be able to drill down or drill out from summaries to details.
Information should also be presented efficiently via effective graphical displays. Simple tabular formats have been replaced by charts, graphs, scatterplots, heat maps, and geographic maps, to name a few.
When looking at data, it should be easy to sort, pivot, or otherwise organize and move around, so that users can interactively look at different perspectives.
Data presentation can be broken down into four major categories: ad hoc reports, pre-defined layouts, what-if scenario generators, predictive models.
The business intelligence community is made up of (i) business decision makers who drive action; (ii) power users, who are the interface between the decision makers and IT; (iii) and finally IT developers and architects. When setting up a business intelligence platform, one needs to set up roles for each type of community member. In addition, members will create their own reports and dashboards, and must have ways to share this with the rest of the community. The community will also be broken down by subject area – for example Marketing, Operations, Sales, Accounts Payable, etc.
Under it all, there will exist two platforms: one for data and one for presentation. Both these platforms are designed to be scalable, fast, secure, and flexible.
The data platform is usually a subject-area data mart, designed for easy access by the presentation tools. The data mart is typically fed from an enterprise data warehouse.
The BI platform is supported by one or more BI software products. In addition to providing all the elements of presentation, BI tools also provide a layer of abstraction between the data layer and the presentation layer.
Any business intelligence implementation should use this high-level overview as a starting point to develop a roadmap for building a robust and scalable BI system.