One may think, that the above is obvious- but it definitely is not. As the saying goes “Ignorance is bliss”, and so a lot of large scale hype can be gathered using ‘silver bullets’. Big words and catchy phrases sets the ball rolling for a number of marketing hypes, and I guess that is just the case even with “Analytics”- today’s new silver bullet. What is more, “Analytics works on ‘Big Data’” which is another favorite of the Brotherhood-of-Catchy-Phrases! I am trying not to be skeptical about the value and contribution of both of these to the world of business, but well both of these are now being overtly misused and misrepresented. Sometime back I was having a conversation with a friend of mine who happens to work in “analytics” for a major and over-sized Indian company, and it went like this:
Me: What’s up? What are You doing these days?
Friend: I am into Analytics. It is “THE” big thing now. [Grins with a big smile.]
Me: Ah Great! So what are you doing in analytics?
Friend: We solve business problems using Analytics for all sorts of companies.
(Now I am interested. Business problems- oh yes, that is what I want to solve as well being a Product Manager)
Me: So what kind of business problems?
Friend: Ah those analytics ones?
Me: What analytics ones? [I am perplexed] So, what exactly do You do?
Friend: Well you see, the customers sends us data and asks a number of questions, and we analyse the data and give reports on the same!
Me: That is what regular data analysis is about! What or where is the “analytics” ?
Friend: Hey that is analytics!
… and I stop myself from asking more questions, knowing the fate and direction of the conversation.
I cannot blame my friend but as a matter of fact, this is what ( I am sure) a lot of people would find when a bit of investigation is done into this new buzz word called “Analytics”. Mindless marketing promotions are the biggest problem creators in today’s world.
Analytics is a means to an end, and not an end in itself. How one applies logic to make sense of data is the key benefit the organization can bring for its customer. The key to data analysis can be summarized in four elementary steps: 1) Define the problem 2) Disassemble the data available 3) Evaluate the data within the environment 4)decide on the solution to the problem.
The biggest challenge is in fact in the “definition of the problem” to be solved. Reducing churn for example is a definite problem to solve. Increasing customer stickiness is a problem to solve. Maximizing the use of available inventory is a problem to solve; and there are hoards of other challenges that can be solved, but the first key is to identify what “needs” to be solved.
The effective way to implement data analysis using analytical tools has been well laid by Thomas H. Davenport when he summarized the necessary steps as:
- Creating and championing the cause of problem solving using analytical solutions from the “top”. The business owners need to be convinced of the problems to be solved first – the how to solve can formulated by other experts.
- Creating a single team who works on the data. One can call it the “analytics” team, who know how to make sense of data using data analysis techniques and algorithms.
- Identification of the “what to solve”
- Having metrics to measure the quantified benefits of a program
- Using the “right” technology.
… and using the right technology can include hoards of ways which would be specific to the organization who is trying to solve the business problem. The final piece of the puzzle could be a good data visualization tool—but that is not the end of it. This brings me to having observed a large number of solutions out there in the market which are essentially data visualization tools but with the tag lines of “Business Intelligence tools” or currently as they say, “Analytics” tools. You wish they would solve the business problems without the feed of the domain. Well, the argument is, “it is obvious to know the domain”. Guess what- knowing the domain in and out in itself is the first challenge, and therefore it becomes far more important to first define the problem to be solved.
What is more, it is time to understand what would be the output of such use of “analytics”. More often than not, the output is provided in the form of a multidimensional dashboard which one would have to first understand, learn the inputs and then try and make sense of what is being shown. Essentially, the output is more of a dashboard which one has to understand first and then try to decipher to make meaningful sense out of it. Now that itself is the problem. What is the use of such “analytics” when one has to give efforts for finding meaningful sense? It is more like giving a GMAT or some B-School entrance examination where a bunch of data interpretation questions are asked to the test-takers. It does not make sense in business, unless the “exact business critical information” is upright visible. That is the INSIGHT that is required for the business executives, and being able to show insights out of underlying data requires thorough understanding of the domain and hence the business problem to be solved. Hence analytics is NOT about a fancy multi-dimensional dashboard representation of a huge amount of underlying data.
Just before I conclude, here is an example. Imagine a Graph is showing a downward trend of sales. This is a concern for the executives. But what is the “analytics” in this?? The graph can be extrapolated with prediction algorithms to show when sales are going to hit rock-bottom. That is also about the visualization and how good the visual tool is. Again, where is the “analytics”? The true value of “Analytics” is the insight to showcase “why” sales plummeted and the consequence of the “sales hitting rock bottom” as predicted by the scary graph. Since the interpretation of the graph is left onto the on-looker – the executive in this case, the real value of the INSIGHT is left open to interpretation. That is what “analytics” needs to be used for- to stop speculations and individual interpretations and provide the real INSIGHT into the state of affairs.
Since I don’t claim to be an expert all rounder like a few other people we may know, in the following posts I would try to share some thoughts on “Bigness” of Big Data!