We, at Analytics India Magazine, recently did a survey of 50 analytics leaders in India on their Business outlook for the year to come. The results were not surprising at all; almost all of them are confident that the demand of Analytics at their organization will increase in next 12 months; all of them plan to increase their analytics workforce.
An interesting outcome of this survey was that almost 3 out of 5 leaders believe that their customers have little knowledge of analytics and this posed as a top challenge for non-adoption of analytics. This came as a close second from “Unavailability of Analytics Talent” as a top challenge.
I have little doubt in my mind that the end consumers of analytics are a misinformed bunch. More so, I believe that it is we as an industry that is to blame for this. In my mind couple of things have happened:
1. Heavily “jargon-ized” industry
Analytics is one of the most jargon heavy areas recently; “Business intelligence”, “Data Science”, “decision science”, “machine learning”, “Big Data”, “artificial intelligence”. It’s hard for even a seasoned analytics professional to distinguish between these terms, leave aside the end consumer of analytics. I recently published an article on how Big Data is not analytics.
Jargons help in initial selling by playing on naivety of end consumer, especially where heavy consulting is involved rather than actual implementation.
2. Exaggerated misinformation
I regularly come across content over Internet around how big data can eradicate poverty, market bubbles, cancer, diseases; predict earthquakes, future, crime, super bowl etc. Obviously, statistics and intelligent algorithms are being used for decades in various areas and analytics is more of an industrialization of these techniques. Analytics is being utilized in various areas with varying degree of success and the usage will only increase; but most of these content is just an exaggerated misinformation without much of the crux.
3. ROI from analytics cannot be fully established
Let’s say I pitch a basic regression model to a customer and explain the benefits of doing so. But, maybe there is no model; maybe there is no correlation at all. I’ll come to know that once I start working on the model itself. This is just a simple example of why it’s hard to pin point a confident result from the very beginning. Unlike an IT project, where we can freeze on a specific output (application, database etc), analytics is as much about doing analytics as the final results.
4. Standards & processes not prevalent
IT industry, during its maturing phase, gave rise to various standards and processes in almost all areas – development, testing, collaboration, maintenance etc. SDLC is one such framework that is an industry standard and everyone in the eco-system abide by it. Analytics has yet to evolve any such standards. Even standards around how do we store, deploy, share for re-use analytics models/ algorithms is still to evolve.
5. A fast paced industry
Analytics is an extremely faced paced industry, what is novel now is outdated in 1-2 years. The pace of innovation and buzz creation outruns adoption. An example being social media analytics, I don’t think there is complete adoption of social media analytics among enterprises, but there is not much buzz anymore.
Given all the shortcomings of the industry, I would believe that the best way to adopt analytics is a continuous experimentation. Persistence is the key; early results may at times even be discouraging. A right implementation of analytics is a source of sustained competitive advantage for organizations; I say ‘sustained’ because not many get it right.
I would also re-iterate from an earlier article that as an industry we need to make analytics ‘unpretentious’ for our customers and not blame them for non-adoption.
Try deep learning using MATLAB