Hang around long enough at a business school that trains both MBA’s and PhD’s and you will hear a joke thought to capture the main difference between the two: an MBA can make the best decision among a set of bad choices, while a PhD will prove that you never really had a choice in the first place.
As saucy as the punch line may be, it also provides a good short answer for the difference between analysis and analytics, after you alter the punch line somewhat.
In brief, analysis seeks to quantify the available choices, and select the best from among them. Analytics attempts to discover all the choices that could be available, especially the ones you did not even know you had.
The longer answer is even more instructive. Examining how analysis and analytics come about reveals some worthwhile principles which we should observe when designing analytics software.
Same root word, different roots
We hear it increasingly from business users: “analytics is just like analysis, but with too much data and not enough results.” It is easy to understand such frustration. Yet we should not allow this sentiment to become a punch line that concludes most conversations about analytics.
Analysis refers to the set of quantitative techniques that both defines a business strategy and renders it operational. Properly done, analysis leads to an outcome of a business decision, along with a framework for allocation of resources, quantification of risk, and establishment of control systems.
The aim of crafting business strategy is of course to create competitive advantage for the company. Since most competitors within the same industry produce similar goods to one another, the best decisions informed by the best plans determine the marketplace winners. For analysts, competitive advantage is achieved through determining the most valid business plan, at the core of which is doing something different from everyone else. The amount by which that difference is meaningful to the market is what we refer to as value, and is the ultimate measure of a successful business strategy
Analytics, on the other hand, is a set of quantitative techniques borrowed from scientific research where the outcome is new knowledge that may inform a decision, but does not prefigure its execution like analysis does. The new knowledge may lead to insight, improved performance, or some otherwise unknown capability whose discovery often creates an immediate competitive advantage. Thus, analytics represents an evolution of the field of action research to an environment of big data.
An important concept that further separates the two is that although analytics makes use of scientific methods, the primary outcome is relevance, not validity. In practical terms, analytics results must be immediately convincing and implementable by all stakeholders.
Competitive advantage is achieved from knowing something that competitors do not know and thereby doing something they cannot do, because they lack the same data. Value in analytics is somewhat more complicated, tending frequently to adhere to the dynamics of intangible assets.
Design principles for analytics solutions
1. The user must determine what data to ignore
Data available for business analysis is often incomplete and typically must be supplemented by additional data in order to yield valid results. Analytics data, on the other hand, is overabundant and needs to be reduced, compressed, or restructured in order to become relevant. In any case, the user will need to ignore the majority of the available data. Analytics solutions should include features that enable sound user judgment on what data to ignore.
2. Tests of findings are necessary to move forward
Analytics is characterized by both discovery and testing. Yet when analytics consists of only discovery, the results are no fundamentally different than gut feelings. Analytics solutions need to provide tests of sufficient rigor so as to transform discoveries into business knowledge, and not reinforce prevailing intuitions.
3. Every step forward involves simplification
Analysis is a process of elaboration and consolidation that leads to business decisions based on qualitative differences determined by the results. Analytics is a process of simplification that leads to new knowledge based on testing information discovered during the process. Each iterative step in an analytics solution should therefore move toward greater simplicity rather than elaboration.
4. Steps must be retraceable
In the spreadsheet environment commonly used in analysis, steps are not necessarily retraceable. For example, data cells can be altered without a subsequent ability to audit, thereby compromising data integrity. Analytics solutions need to log all user activity, allowing the user to retrace all steps performed.
5. Good enough is good enough
Good enough means knowing when to stop, and is achieved when the validity of the results matches their relevance. Therefore, an analytics solution should be designed with the primary aim of the user obtaining relevant results.