The Coming Revolution In Business Analytics

Sometimes, so many pieces seem to line up, I feel obligated to make a call and declare "big changes are coming".

Such is the case with a topic I've come to call business analytics.  Other people might call it decision support, data warehousing, data marts, or perhaps something else.

At its highest level, it's using analytic results to make better business decisions. 

And I think it's a topic that many of us will be drawn into at one level or another -- if we're not already.

To Begin With

Perhaps the most articulate spokesperson on the business agenda is Thomas Davenport.

His original book "Competing on Analytics: The New Science of Winning" laid the conceptual framework for executives everywhere.  This was recently followed up by "Analytics At Work: Smarter Decisions, Better Results" which went even deeper in exploring the dynamics behind this new competitive weapon.

I've had the pleasure of following Tom for a while, and recently meeting him in person for an in-depth conversation.  He's very smart, personable, articulate and extremely passionate.  I'd like to get him in front of EMC as a management speaker at some point, for example.  You can follow his blog here.

His core premise is devastatingly simple: better get good at this new business weapon, because your competition will.  Based on reading his material -- and my own interactions with customers -- I can't disagree.

But exactly *how* you best do this -- especially from an IT perspective -- is up for a serious debate, especially in consideration of some of the macro trends that have recently heated up.

Traditional Thinking Is Up For Change -- In So Many Ways!

As we peel the onion back, it's easy to see how so many fundamental assumptions around this practice will need to evolve if we are to achieve the stated goal: competing on analytics.

Traditional thinking is largely around running and optimizing the same analytics reports and queries, over and over again.  Do exactly what you did yesterday, only do it a little faster and a little cheaper.

Well, that doesn't ever go away, but it's clear that the new value now moves to answering entirely new questions that haven't been asked before, rather than answering the same questions over and over.

The corollary to this is "speed matters".  Gone are the days of batch reports; business will value actionable information far more when delivered in near-real-time vs. traditional analysis that is days, weeks or even months old.

Traditional thinking has always been around The Data Warehouse, with that sanitized, cleansed, rationalized, master-data-managed, one-version-of-the-truth picture of the business that we as technologists inherently long for.

Well, the trends seem to be going in exactly the opposite direction, as far as I can see.

In the new view, a data warehouse is just one of many potential interesting data feeds internally and externally.  Indeed, new data feeds are becoming available in the marketplace at a dizzying pace (I'm thinking about social feeds, or location-aware feeds -- as an example), and there's inherent business value in being able to exploit and correlate the very latest relevant data feeds as they become available. 

No time for the months-long project of integrating them into The Data Warehouse.

Data quality and the whole topic of master data management is also up-for-grabs as well.  Good business analytics people know how to deal with imperfect data, and create their own subjective and pragmatic views of data meaning that directly correlates with the task  at hand.

It's far more likely that there will be many versions of information "truth" in large competitive organizations, rather than just one official one that's controlled by IT.  Sure, there will be traditional use cases where it makes sense for everyone to work off the same data, but that won't be where the new value will be created, as far as I can see.

Got Data Marts?

Often when I talk to enterprise IT people, they get very frustrated that everyone has their own data mart that they use for reporting and business analytics.  Indeed, there are many projects around "rationalizing" all the different data puddles, and "standardizing" on a few data sources and associated tools, often in the interest of "saving money".

I think -- for the large part -- this might be completely misguided thinking.

I have been a business user of several of those annoying data marts over the duration of my career.  I have used them to make timely decisions that ultimately made money for the company I was working for.  At the time, I fought with IT people who were very disapproving of what I was doing.

Too bad, I thought.  My job is to make money for the company, IT's job is to help me do that.

So, if you agree with that general sentiment, how might that reflect itself in a new way of thinking?

Imagine self-service data marts, ideally running on optimized virtual environments.  I, the business user, get to select my preferred data sources, select my preferred tools and drive my own analytics.  IT provides the catalog of data sources, tools and pooled / optimized resources for me to you. 

As a business user, I am more than happy to pay for this IT service.  I get what I want, when I want it, and I can change my mind as the business changes.  Just please don't tell me how to do my job :-)

Indeed, as we talk about private clouds and things like Vblocks, one of the more popular use cases I've noticed is creating self-service business analytics environments for cranky business users who don't like the speed and flexibility of their current IT-provided environments.

Cost savings comes from virtualization of storage, server and network.  Speed and flexibility comes from putting the resources directly in the hands of the business users with no questions asked.

And There Are Technology Disruptions As Well

So much of what we do in this space at an enabling technology level is also up for grabs, if you follow this stuff.  Spinning disks are being replaced by increasingly cost-effective flash drives.  Large terabyte-class memory spaces means that more data can be directly processed without the usual I/O considerations.

Even data management technology is undergoing a revolution. 

The classic old guard of Oracle, DB2 and Teradata are already being seriously challenged by the newer columnar players like GreenPlum, ParAccell and Vertica.  And, beyond that, there's a whole class of technology built on "big data" principles with near-infinite scale-out principles -- like Hadoop and its enterprise equivalent, Cloudera.

Putting It All Together

I'm starting to see a virtuous cycle start to form, and I expect it to be in full force within the next few years:

• Businesses are demanding a new class of analytics as a competitive weapon.  They want to ask new questions, from multiple data sources, and they ultimately value speed and flexibility over traditional concerns such as cost and efficiency.
   
• Knowledgeable business analytics users want IT resources that they control, and not IT.  They want to choose their data sources, their tools and their processes.  And what they want on one day won't be what they want on the next -- expect it.
   
• The enabling technologies -- storage, memory, CPU, data management layers -- all are now showing the potential of delivering order-of-magnitudes improvement in speed and flexibility. 

And, you have to ask yourself -- what happens when entirely new business questions can get answered far faster (and far cheaper) than ever before?

The answer is easy: people will ask more -- and more interesting -- business questions.

And that, ultimately, is what competing through business analytics may ultimately be all about.

Comments

[josephmartins]

I've followed Davenport for years now and believe he's spot on most of the time.

Technology aside, I've had dozens of conversations about analytics with executives who, unfortunately, simply could care less. Frankly, Chuck, some of them don't want to change the way they're doing things or risk revealing, perhaps, that what they've been doing is half-baked wasteful nonsense. Others don't want to make the investment because they can't wrap their minds around the long-term value of analytics.

Companies such as Primary Intelligence (no affiliation) have been delivering outstanding value-add sales and marketing analytics and insight for years now. How many people know that such companies exist? Relatively few. I've been hopeful that, with acqusitions such as IBM/SPSS, the 900 pound gorillas of IT might do a better job of educating the market. Time will tell.

Yes, the revolution is coming. Just more slowly than we'd like. If you do convince EMC to host Devenport, it'd be interesting to see how many of the attendees take his advice seriously and welcome analytics into their processes.

[Chuck Hollis]

Hi Joseph

I think it's a completely relevant topic to explore "how do new ideas and practices become prevalent in the business world?" with proficient use of business analytics right up there with globalization, diversity, sustainability, six sigma et. al.

I tend to have some half-baked thoughts around the processes at work that cause us to use and adopt new ways of doing business.

Maybe worth exploring, or not ...

-- Chuck

[Chuck Hollis]

Chuck here ...

I'm getting a lot of back-channel discussion around what appears to be a dissing of MDM and related topics.

That wasn't my intent, really.

Upon reconsideration, I probably should have said that having nice, well-organized and sanitized data can only get you so far -- at some point, real business value comes from mastering the other kind :-)

The real discussion might be -- what are the limits to MDM in this new world? How much should be shared views of data, and how much are independent and context-dependent?

Now *that* would be an interesting discussion, no?

-- Chuck

[ajay]

Making Analytics sexy doesn’t make it easier to implement! And this is where the challenge lies in Analytics. Striking headlines make for easy copy but don’t do wonders for executing analytic intent within a corporation.

Unfortunately executing analytics is hard work and demands the coming together of business skills, advanced statistical knowledge and technology capabilities.

http://blog.cequitysolutions.com/Customer-Management-blog/bid/12403/Analytics-is-sexy-what-about-making-it-work