They -- rightly so -- think there might be a powerful new tool at hand, and want to find out more. I consider myself privileged to speak to them.
The basics of my discussion are fairly straightforward: here are the core concepts, here are why they are important, this is what an organizational model might look like, here are the key skills and roles, here are some examples of how leaders like you went about moving forward, here are some interesting case studies, and so on.
Normal transformational fare, in my world.
Where it routinely gets sticky is when people struggle is between unlearning the familiar world of traditional BI and gaining an appreciation for the new world of data science. I struggle to find an effective and repeatable technique -- I have yet to crack the code here.
But it's important to do. Unless business leaders realize what's now possible with a fresh approach, they'll only end up incrementally improving what they're already doing today, as opposed to taking that breathtaking quantum leap into a new realm.
If that's the case, why bother? So I consider myself less than successful unless I'm able to un-stick a significant portion of the audience from their old perceptions, and wrap them around a greater appreciation of newer concepts.
I'm looking for that "aha!" moment.
The Challenge Of Unlearning
This is not the first time I've struggled with the inherent challenges associated with unlearning.
Many years ago when I was championing social media proficiency here at EMC, I found myself comparing the proposed social proficiency approach with more traditional corporate communication and collaboration mechanisms: email, doc repositories, intranets, etc.
That ended up being a sort of conceptual tar pit that took significant effort to claw my way out of.
The benefits of having a thriving internal social community is hard to compare and contrast with, say, corporate email -- you tend miss the essence of the new concept because the starting frame of reference is way too limiting.
At a more tactical level, every time EMC introduces a new technology that isn't an incremental variation on an existing theme, it usually takes a considerable amount of time and effort to communicate what's different. One example from storage that you might be familiar with is Atmos. And, of course, everyone starts by comparing the new thing to something they already know quite well, e.g. NAS in this case.
Some people are quite adept at freeing their minds from established baselines, conceptualizing the new concept, and then working backwards. Many more just can't do that easily. And I've learned that -- unless you can facilitate this -- it's going to be a long slog indeed before your ideas gains any traction.
This is turning out to be the case when I speak to business leaders about achieving big data analytics proficiency.
Stepping back a bit, there are three kinds of customers when it comes to this topic.
The first inherently gets it: they're doing it, they see the value, and they want to do much much more with the resources available. The discussion immediately veers into products, technologies, workflows, use cases, etc. As a technology vendor, we're on very familiar ground.
The second category doesn't have much of a starting point: they're not doing much with analytics or reporting, they don't generate or participate in large information flows, and so on. Other than a bit of general awareness, there's not much we can bring to the table for them.
The third category is where I'm mostly focused: progressive businesses who have a decent foundation, and an appetite to potentially do more.
So, how do you go about "unlearning" someone when discussing big data analytics?
Approach #1 -- Compare And Contrast
The most popular approach is use a side-by-side chart, comparing and contrasting old-school with new-school. In this case, you'll always see some familiar elements when comparing traditional BI and data warehousing with the somewhat newer practices around data science.
Limited internal sources vs. a broad variety of internal and external information sources.
Clean, sanitized and managed data vs. messy, unfiltered and fresh data.
Focus on "what happened" vs. predicting "what will likely happen"
And so on.
That sort of side-by-side comparison works -- up to a point -- but it's usually not powerful enough to shake people free of the current mental framework, and adopt a new one.
Approach #2 -- Before And After
More recently, we've collected several customer examples of business process performance "before" (using traditional BI techniques) and "after" (using predictive analytics techniques). Surprise, surprise -- the big data enabled business processes performed much better than the other ones.
The outsized results leads the audience to believe that the root cause was not "doing more of the same", obviously something fundamental and substantial had changed, which leads to a natural intellectual curiousity as to what was different, and how they could do that in their own organizations.
Better -- you're more likely to get to that "aha!" moment, but the approach is still somewhat cumbersome and not always effective in larger settings.
Approach #3 -- A Star Is Born
Just to throw a little gasoline on the fire, I share two or three stories of traditional businesses that got so good at analytics and creating useful information products, they created and spun off an entirely new business unit to pursue the opportunity.
In some cases, the resultant entity is now worth more than the business that created it.
To my way of thinking, if that sort of story doesn't get your entrepeneurial juices flowing, nothing will :)
Approach #4 -- Observed Cultural Changes In Mindset
If none of the above are working, I bring out my heaviest ammunition: examples of how mindsets and behaviors are fundamentally different in organizations that are proficient in big data analytics.
For example, if you want to really understand France (or India or Japan or anywhere else), it's useful to give examples on how people in that country tend look at things differently than you might. You start to realize that maybe it's your perspective that's holding you back :)
One important observed mindshift is the emergence of a complete and utter confidence in black-box predictive models that have been shown to be statistically relevant, but aren't fully understood as to why they work so darn well.
A more traditional mindset usually won't accept any answer unless there's a thorough understanding of "why". In the new world of big data analytics, that's not the priority -- just capitalizing on the predictive power :)
My colleague Bill Schmarzo covers the topic well in his post on "The Death Of Why" -- as long as you have a predictive model that works, it ends up being much less important to understand "why".
Another key perspective shift appears the "wide net" phenomenon. A traditional mindset leads you to named data sets of interest, a list of good reasons why you might be interested, and attempting to fit them to well-understood analytical models.
Contrast this with the newer approaches (for example, the new work around MINE - maximal information-based nonparametric exploration) where the goal is to simply sift massive amounts of information and find "interesting" areas to drill in further -- far in advance of any preconceived model or potential assumptions.
To use an inexact analogy, it appears to be the difference between processing every cubic yard of sand on the beach looking for metal -- and using a metal detector to find something interesting -- even if you aren't quite sure what it is until you dig it up :)
Compare that with the normal practice of "have a theory, let's go find some data" which becomes inverted to "found something potentially interesting, wonder what it might mean?"
There are other relevant mindset changes, but I think you get the idea.
By highlighting how advanced practioners think about things differently than you might, the "unlearning" process often gets accelerated.
Or so I've found.
The Challenge Of Changing Perspectives
There are so many great ideas and technologies out there for us to go apply to make the world a better place. Our propensity to innovate has greatly exceeded our ability to apply and consume.
When it comes to big data analytics, we've got an excellent example of this maxim.
The technology is there. The results are widely known. The broad applicability has been established.
All we have to do is change a few mindsets :)