The premise -- at a high level -- is easy to understand: the new-found power of predictive analytics has the potential to change the way you do business.
Clearly there's a transformation at hand.
The starting point: a familiar world of historical reporting on events in the past. The goal state: the ability to better predict (and capitalize on) likely future outcomes.
And getting from Point A to Point B is rarely as seamless and pleasant as you read about :)
To that end, I'm actively studying transformational stories about how organizations are learning to change the way they do business around big data analytics.
And, as it turns out, EMC Corporation is an excellent example of just such an organization.
Our Starting Point
The challenge (or opportunity?) is that there are multiple parts of our business where it has become obvious that things could be radically better if we learned how to leverage the power of big data analytics.
But change is hard, isn't it?
There are established ways of doing things: established roles, established behaviors and established perspectives.
The supporting technology is proving to be relatively easy. Getting people to change the way they do things -- now that's hard.
And therein lies our story.
Meet The EMC TCE Group
TCE is YAA (yet another acronym) at EMC, and refers to the Total Customer Experience. I suppose you could call them "customer quality", but their mandate and scope far exceeds what you traditionally might find at an IT vendor.
I've written about them before, and -- if you're interested -- the back story provides useful context as to why they're embracing big data analytics as one of the very first use cases at EMC.
If you want the bullets, here they are:
- EMC is incredibly focused on customers -- part of our core DNA.
- The TCE group is the primary organizational feedback mechanism to drive operational change within the EMC organization based on VOC (voice of customer).
- They have ready access to multiple, rich data sets that correlate around customer experiences with our products and services.
- They have a sufficiently high level of executive sponsorship and associated funding within EMC.- They are reasonably proficient with statistics and analytics.
- They are passionate about what they do, and are willing to work hard for change.
- The changes they drive have historically made major impacts in EMC customer experience, and thus are a major lever for our business model.
Net-net: if you were looking around for a perfect set of textbook conditions to introduce big data analytics, it'd be hard to find a better set of circumstances.
Should be easy, right?
Jennifer Bodzinski heads up the analytics team in the TCE group, and she puts it simply.
She's one of the earliest use cases for the BI-as-a-service our EMC IT team is standing up. For her, the benefits are simple: data sets are now far easier to source, consume and experiment with.
Like 10x-100x easier and faster for her analytics team.
It's not so much that existing work is getting done faster, it's such a dramatic improvement that entirely new things are possible in her world -- and, by extrapolation -- across EMC itself.
So, how does this "change-the-game" capability change her world -- and EMC's?
A Huge Help In Overcoming Organizational Resistance
When her team approaches an internal group with a "I think we've got a serious problem" event, things don't always go so well.
She summarizes her historical engagement as a three-phase process.
- First, there's a strong denial that any serious problem might exist. This is known as the "I haven't heard of any problems" response, usually followed with the "it's not my problem".
- Second, there's an argument about the data -- where it was sourced, how it was interpreted, etc. Remember, we're dealing with very smart people here, and they know how to mount a defense.
- If the first two defensive tactics don't work, frequently the discussion moves to attacking the credibility and the motivations of the messenger. Unfortunate, but that's what people do occasionally.
Given their new capabilities by using big data analytics techniques, she and her team now show up to meetings with 10x the firepower before -- an overwhelming and incontrovertible amount of correlated data with unarguable conclusions.
The net result? The three phases of resistance can now take minutes, instead of months.
Keep in mind, that sort of analytical firepower can be aimed at either the people chartered with owning the resolution -- or their management if needed.
She has plenty of examples where this has worked well. She cited one situation where a routine quality issue that usually took 11 months from the first indications to resolution -- to under a month of closed-loop cycle time. Sure -- dramatically better -- but still room for improvement.
Empowering Process Owners
They are raising their hands, and very motivated to get into the data and the methodologies used.
Jennifer's team now has the ability to share their "analytical sandboxes" with other business process owners who are so motivated.
As a result, a very nice and elegant segmentation of responsibilities is starting to take shape. Jennifer's analytical team is responsible for correlation, the business process owners are becoming responsible for causation.
Put differently, the TCE team uses big data analytics and correlates multiple data sets to identify a problem area, suggest a root cause, and quantify the impact to the customer as well as EMC. With increasing frequency, there's now a clean hand-off to the business process owner to dig deeper to understand the problem -- and possible resolutions -- at a deep and granular level we've never had before.
If you're interested, the primary stack for this particular use case is the Quality Lifecycle Analysis Framework capability from SAS, running on a Greenplum UAP environment.
From my personal perspective, that empowerment of business process owners is nothing but absolute sweet goodness.
Predicting Potential Customer Events Before They Happen
Over a decade ago, we were sort of proud that our storage arrays monitored their internal health, and would call for help proactively before there was a customer-visible problem. Progressive capabilities in their time, but we're starting to play in a whole new world these days.
In essence, we're quickly moving from a reactive model to a predictive one.
For starters, take a quick look at the data sets that Jennifer's team is starting to mash up.
- Historical quality data multiple sources around products, components, even people and processes.
- The service history of customers, both individually and collectively.
- Costs associated with customer service -- both to EMC and our customers.
- Customer experience data gathered from a variety of internal and external sources.
- Supply chain information: both historical and near-realtime.
- Manufacturing and warehousing data; inventory, in-factory qual processes, etc.
- Engineering databases
- And a few other interesting sources as well :)
Her team is starting to come up with a variety of predictive models that can (a) model how likely there will be a particular issue in the future, (b) the likely impact and associated costs from the customer's perspective, (c) the likely impact and costs associated from EMC's perspective, and (d) modelling different scenarios for mitigation of the issue: costs, risks, etc.
All of this potentially many months before there's even a whiff of smoke in the air.
More data sources means better predictive models which means more time and precision on how precisely to react best.
Now, think about the impact of this for a moment. It's not an overstatement to say that this sort of capability changes the game -- not only for us, but for our customers. Sure, it's early days in making this sort of capability uniform across EMC's businesses, but you can imagine the enthusiasm we have for big data analytics in our own business.
Empowering Our Customers And Partners
Now -- let's take the next logical step.
Previously, I mentioned that we were starting to share "analytical sandboxes" with internal stakeholders. Here's the data, here's the tools, here's the models and methodology -- have fun!
Imagine if we made the same capabilities available to our customers and partners?
Many of our customers -- especially the larger ones -- take vendor supply chain management very seriously indeed. Right now, we're usually emailing each other summary powerpoints back and forth based on ridiculously limited data sets vs. the tantalizing potential of sharing predictive models built on rich data sets -- and discussing better ways to do things.
More and more IT organizations are starting to look like service providers. They're going to care just as much as we do about keeping their customers happy, and ultimately predicting potential problems vs. simply trying to react as fast as possible. Imagine if we could give them a huge leg up in doing just that.
That would be truly transformational, wouldn't it?
Advice From The Experts
I asked the team what advice they'd give to anyone in a similar situation. They thought about it for a while, and came back to me with a very useful list.
First, there's no substitute for strong and persistent executive sponsorship. The team is driving change across the organization, and it's essential to have one or more very powerful sponsors behind you.
Second, take your business case very seriously. Make it rich and engaging enough that there's something for everyone: hard cost savings for the bean counters, soft measurements for people who think that way, and don't forget to appeal to the innovators and change agents out there.
Third, think in terms of multiple small wins vs. a big one. As the TCE team has progressed, they've made dozens of incremental steps, each with a well-documented and well-understood set of business benefits. Resist the temptation to go big.
Fourth, position the effort as "building strategic capabilities" vs. solving specific tactical problems in the business. Everyone has their pet problem to go solve, and it's easy to get sidetracked into endless tactical goals (like managing headcount) vs. strategic ones (like reinventing the core business processes).
Fifth, get agreement from the IT team early as to your business goals and your working model, especially if you're working with a traditional enterprise IT team.
There seems to be this innate tendency for traditional IT people to tell business people what they can and can't have, demand precise requirements, become extremely concerned about operational details, and so on. All perfectly understandable, but this is an entirely different style of IT as compared to a traditional environment.
Working with big data analytics is more about R+D than running a production SAP instance: you're discovering new stuff, and the IT environment is more along the lines of what you'd find in a research environment vs. a production environment.
And, finally, try to be patient yet persistent. Don’t take things personally. You're driving change, and change at scale is never simple, easy or quick.
Wise words indeed :)
The TCE group at EMC is only one part of our business that has the potential to be radically transformed by using big data predictive analytical models. An important part, to be sure, but only one component in an incredibly large and complex organizational machine.
There are at least a half-dozen likely targets there in our company that I'm aware of. From those handful of starting points, I'd expect many dozens more to follow, and -- one day -- we're going to wake up and like will be very different at EMC. I'm guessing this will happen in a reasonably short timeframe.
My goal? Keep you up to date on how we're transforming our business using big data analytics.
If you're a long time reader, you'll probably remember how I did something very similar with corporate social media proficiency. It ended up fundamentally changing the DNA of our company.
This should be even bigger ...