The views expressed on this blog are my own and do not necessarily reflect the views of any organization that I have worked with.
As is usually done in the world of Analytics, we use (not necessarily depend on) past data to predict the future. So let us first have a quick look at the past and present of analytics. Now, all of us have indulged in different facets of analytics while Business Intelligence & Analytics grew exponentially over the past 20 years. Clearly, we all will have different perspectives of the past. From my eyes, I have seen the progression from Database based to Data Warehouse based analytics, from Strategic to Strategic and Operational to a hybrid where the distinction might matter no more. The sources of data have increased from ERP, home grown systems to Cloud based data, Social data, Public data and Machine generated data. Then the third aspect of this growth “What can be done with all this data and technology that is available to us?”. Enter Visualizations, Predictive Analytics and Form factor based analytics (mobile, tablets etc). With the need to do much more with all that data that is now available, we saw technology strides like Big Data and Business Intelligence appliances. It almost looks like we are at a brink where we have everything we can do with all the data we have, and now it is just a matter of doing more with the data, techniques, visualization and technology available.
And that is probably where we will take the next few steps: ‘Rise of Analytical Applications’. Clearly more and more organizations are already providing out of box analytical apps for CRM, HCM and other domains. I can see a lot of handshakes happening in the industry where organizations tie up to enrich out-of- box Analytical Applications with data from other software vendors, social infrastructures and public domain. While there is a huge upside to this – more data available in an integrated manner to various business users – a paradox seem to evolve here. On one hand, we have an explosion of data sources that can be integrated, coupled with multiple visualizations and techniques being available; the other – the form factor for display seems to get smaller ( desktop to tablets to phone to smart watches). Also the problem that availability of more analytics could lead to ‘Analysis Paralysis’ and ‘Information overload with conflicts’ persists.
Would then the second step be where the business user says “Analyze the analysis for me, please!!” In other words, “Please do not show me all visualizations. Instead, summarize your findings for me”. Interestingly, this is a business user asking her smartphone or smart watch to give her a one line answer and not paragraphs. And may be that is exactly where Analytics goes next. Three words – ‘Infer’, ‘React’, ‘Suggest’. For years, I have believed that ‘Enterprise Brain’ will follow ‘Human Brain’ in terms of its working. In our day to day life, we gather ‘experience’ (sometimes subconsciously). When we see that a certain experience repeats (like a friend being available most of the time when you need something), we mark it in our mind as a ‘learning’. That friend is marked as ‘dependable’ in our mind. Now we know that certain friends are dependable for certain types of situations. With this learning in our mind, we quickly ‘evaluate’ “whom to approach for a particular situation” when that situation arises. Note how all of this happened subconsciously. This is exactly how Analytical Apps can shape up to convert transactions and day to day analytics into experience, learning and then evaluations. Not very different from current day “Predictive Analytics” – you ask? True, but note how all of this happens subconsciously at an ‘Enterprise level’. We are talking an “Enterprise Intelligence Engine’ infrastructure here that stores experience and learning and creates evaluation / prediction models – all of this subconsciously. The result: three words – “Infer, React, Suggest”. This intelligence engine will then auto ‘infer’ the next time an event happens based on ‘experience’ or ‘react’ based on ‘learning’ or ‘suggest’ based on ‘evaluation/ prediction models’. the result – either your smartphone/ smart watch will summarize the event and possible inferences/ actions, or in some cases, might just inform you of the action that was taken. Note how we are progressing from “Actionable Intelligence” to “Intelligence based Actions”. I know what you are thinking..”Do I loose control to the Intelligence Engine?”, “Do various folks not make different inferences from the same event in real life? In that case how do we ensure that the Enterprise brain also captures similar variations?” Time to discuss the next step in Analytics, I guess.
And that next step is “Bridge the gap between Man and Machine” thereby “allowing Man to trust the Machine (Enterprise Intelligence Engine)”. As in day to day matters, trust is built over time. And more importantly it takes an effort – conscious and unconscious – to build trust. So what would “building trust in Enterprise Intelligence Engine” efforts look like? This is where all three words become significant – ‘Enterprise’, ‘Intelligence’, ‘Engine’. Enterprise is a social body. And to make this society function like one, collaboration will be the key. We are not talking simple messaging services here. We are talking ‘meeting and understanding of minds’. There is ‘Enterprise Psychology’ involved here. We will clearly need tools that help get the minds together and set egos aside while evaluating personal biases for their usefulness or significance . These will participate in driving inferences and actions for analytics. And that leads us to the next word “Intelligence”. Intelligence differs from Knowledge in that Intelligence is what one can do with available knowledge. It is almost like a personal trait. And with a concerted ‘Enterprise thinking’ process build through ‘Enterprise Psychology & Collaboration’ efforts, organizations will have to invent ‘self learning and unlearning, iteration friendly, agile analytic processes’ that help ‘define, create and improve’ the ability to increase content in the ‘Enterprise Intelligence Engine’. And that brings us to the third word ‘Engine’. Being a mechanical engineer myself, it is but natural for me to draw comparisons with engines like the internal combustion engine. Design superiority for an engine is evaluated by many factors, though performance, efficiency and lifetime are key elements. How strong has been the performance of the Enterprise Intelligence Engine? How efficient has it been in various situations? And does it have longevity? Can it absorb changes in market situations and business models? Is it flexible enough to live long? Can it evaluate what experiences are not valid any more? Can it give more weight to new experiences that matter but might have not many events to support them? As organizations build the “Enterprise Brain” that attempts to put the “analytical inferences” on a auto pilot mode, there is one more nuance that needs to be thought through. It is no more just humans generating experiences or making decisions – we have machines too that behave more and more like humans. Enter the next stage of the future of analytics – ‘Understanding Machines’ .
Ever heard of ‘inducting a Machine into the work force’? Well we might now. Imagine replacing one CNC machine with another on the shop floor. Do you not want the new machine to exchange notes with the old machine? Do you not want the old machine to tell the new machine about nuances of that shop floor – and the changes to parameters that are necessary because of that? If one CNC machine experiences sudden vibrations from a nearby forging machine, do you not want the CNC machine to auto correct itself as also let other machines know of the potential impact? Do you not want all machines to meet at lunch time and discuss how they could increase output for the remainder of the shift? Machines and humans have some interesting differences. While one human replaces another in the organization, it is possible to replace a machine with another but still reuse some parts of the old machine like its motor or controller. And not just parts, you could carry forward that old machine’s brain as well. If you have logs and experiences of a machine captured over five years, would you not want that to be added to the new machine – you get the agility and technology of a new machine with the learning of the old machine. And then think about this machine talking to all other machines made by its manufacturer across all shop floors in the world, sharing experiences and learning. My point is organizations will have to consider both ‘Man’and ‘Machine’ in their social and technical efforts while building the “Enterprise Learning Engine”. The next job posting from the ‘Business Intelligence team’ might just be for a “Psychology major who can also talk to machines”.
There are ofcourse many more thoughts in various dimensions regarding the future of Analytics. But let me take a pause and wait for your comments on this blog site/ Linked-in or send your replies to abhaykulkarni@hotmail.com.
***
Σχόλια