Business Intelligence is a phrase that means many things to different organizations, which is why all BI vendors have their own definition. The term was coined in 1958 by Hans Peter Luhn of IBM*, who described it as “the ability to apprehend the interrelationships of presented facts in such a way as to guide action towards a desired goal.” In the ensuing half century, through the evolution of computer technology, BI implementations are myriad, but at their core, all are decision support mechanisms.
Typical BI implementations include reporting and visualization of facts, data both raw and analyzed. Think of it as an interactive presentation layer that allows you to compare and contrast, focus and ask questions – quickly. It won’t typically identify hidden patterns or trends – for that you will need statistics and other advanced analytics. It won’t scrub, prepare and organize your data – for that you will need data integration tools. But when you pull all 3 together, the presentation layer (BI) becomes the moment of truth – it allows you to quickly understand the journey, to understand self and see how everything fits into your financial, industrial and social ecosystem.
Zen and the Art of Decision Support
In yoga class this morning, I was thinking about writing this (multitasking to take one's mind off of creaky joints), when it occurred to me that yoga practice shares many traits with successful business intelligence initiatives.
• Begin in Mountain Pose – You are standing, facing forward, seeing what is immediately in front of you. But how much better can your business strategy be if you also consider past performance?
• Step the right leg back, then the left, to come into Plank – With a top-down view, explore operational and transactional data sources to understand what contributed to both positive and negative outcomes. Are there anomalies present, odd “shapes” in the data?
• Slowly bend arms and lower into crocodile pose – Viewing historical data, do a deep dive to analyze correlations. Did a particular marketing campaign lead to an increase in revenue or new sales? Does the correlation truly imply causation, are there other factors to consider?
• Rest tops of feet on mat and extend arms, coming into upward facing dog – From the historical exploration, are there other data sources that could have contributed to the sales lift? If we are selling outdoor sporting equipment, did the up tick coincide with summer in particular geographies? Suppose we identify a positive correlation between sales of baseball equipment and spring training or between snowboards and an unusually severe winter. What can that tell us about the present?
• Flip feet back down, and push up through thighs and arms into downward facing dog - Armed with an understanding of the past, let's take a look at the present data. Do people buying snowboards from our online store also buying goggles or helmets? Are there bundles we should suggest when only a board is added to the cart?
• Step the left leg forward, grounding the right leg, windmill arms coming into Warrior II. Gaze is over extended left arm. Focus on the future, but continue to analyze current activity. It's now summer and we're opening new stores in Nordic countries in January. Which winter sports are most popular where our retail stores will open? How can we find out and then use that data to optimize supply chain scheduling and inventory levels?
• Step the right leg forward, arms float to sides, returning to Mountain pose – Through a series of yoga poses, one explores space and our own bodies, much like analyzing data. Are there meaningful clusters? Are there gaps in the data needed to optimize outcomes? But always, the practitioner returns to Mountain pose, informed, balanced, and clear. And when BI is done right, that is the return state through each cycle.
The Yin and Yang of BI Phases
Like yoga, BI is a practice. It requires flexibility and regular adjustments as business needs and economic factors shift. When embarking on a first or renewed BI initiative, as with a new yoga pose, it is important to identify your goal. Specifically, what business problem are you trying to solve or what decisions are you trying to improve? Understand who will use the business intelligence generated, how they will use it, and what is their need for currency of data? An executive dashboard may calculate profit and loss based on quarters or years, but line of business managers may need daily or more frequent updates to spot emerging trends to tune operations.
In my years using SAS Business Intelligence tools to analyze software defects, I quickly learned that once I answered my questions with the data, I immediately had additional questions. How did the average time to resolve bugs vary by platform or technology? Was the time to resolve increasing or decreasing after a platform upgrade? BI deployments typically follow this route. Created to improve one aspect of business, when successful, the deployment may be extended to other areas of the business.
This is why typical BI initiatives go through phases. By keeping phases distinct, businesses can use agile development techniques to reduce time-to-value and more quickly deliver on project objectives. Using good project management techniques, clearly define each phase, set the scope (and guard against scope creep), and establish a baseline so that you can measure your return on that investment. Closeout each phase with lessons learned. Were data deficiencies identified in this phase that should be resolved before another phase? Were assets created that can be reused and adapted in another phase or by another group?
One of the greatest lessons about business intelligence initiatives is that they are not static. Business and technology are constantly moving, needs are changing, and a successful BI deployment adapts to that change, remains flexible, and continues to explore data, visualizing it in new ways to gain insights.
* Source: H.P. Luhn. “A Business Intelligence System.” IBM Journal of Research and Development. Vlume 2, Number 4, Page 314 (1958).