The Blog

  • April 24, 2013
  • The New Analytics Wave

    I can’t tell you how many emerging analytics companies have contacted me since January.  Every day it seems there is another company — smelling blood in the proverbial water — wanting to brief me.

    I know why.  Now that Big Data questions have transitioned from “how do we store all this stuff?” and “what’s valuable in this pile?” to “how can we slice and dice this raw material,” everybody wants a piece of the action.

    I’ve been preparing you for this for the last couple of weeks. You may recall my recent post about the relentless doubling of Moore’s Law and a new book, Race Against the Machine that discusses how our world is changing now that we have significant computing power.

    You may also recall my last post about how knowledge is doubling at a ridiculous rate — by 2014 it’ll be doubling every 11 hours, as forecast by IBM. It’s no surprise, then, that analytics companies are as numerous as chocolate bunnies at Easter.

    Here’s the tricky part: A typical early market starts out with messaging about the technology. It lets you do this or that with the data. Later on the messaging includes real quantifiable benefits that you can literally take to the bank in the form of ROI.

    Then the ultimate validation numbers start rolling in and they are usually in the form of double-digit percentage increases and they don’t depend on return on investment because they are so fundamental that they stand on their own.

    From what I see, this isn’t going to happen for this analytics wave. I suspect that there might never be a roll out like that again.

    Analytics have been around for a long time and there have actually been several false starts. There have been times when analytics looked like a sure solution to many company ailments.

    Some analytics vendors did quite well in that era — keep in mind that SAS is a multibillion-dollar company and it is still privately held. Did I mention that it’s about 30 years old?

    Analytics, however, failed to break through because there was always something sexier and easier to understand coming out at the same time.

    Back then, analytics suffered from the reality that you needed a Ph.D. in statistics to really appreciate the stuff. Most companies have very limited supplies of doctoral candidates. They have many more sales people who need leads, so in the showdown between analytics and CRM, analytics always won.

    Then you should consider ROI and what SaaS computing has done to it. SaaS turns big investments into small numbers, so the division calculating savings per dollar invested no longer gets the attention it deserves. That means ROI as a driver just isn’t what it used to be.

    For analytics to be successful this time — and we really need it to be successful — several things have to fall into place.

    It has to be dead solid easy for mere mortals to use analytics. That means each of us needs to up our game. The jobs of tomorrow will be based on the ability to make sense of metrics and probability. That’s not a hard lift.

    We also have to get the messaging down, all the way to the fundamental numbers that define it as worth having — period. We need to do that post haste.

    Most vendors I’ve seen don’t have that picture in mind yet. They talk about their products like they are ends in themselves, instead of being means to the ends of profit and cost abatement. The best way to get the messaging right is to quit talking about ROI and begin talking about metrics — specifically the metrics that are germane to my business processes.

    By implication that also means getting comfortable enough with the idea of metrics — that each of us can easily come up with our own now and then because we are the ones closest to the reality that needs to be described.

    Again, that means we both need to be able to punch up an analysis, but we have to be fearless enough to take command and roll our own. Analytics won’t be mainstream until you can do this.

    I think some emerging vendors are going to do that and then we’ll have a horse race.

    This reminds me of the early days of SaaS computing and CRM. Back then there were multiple companies bringing SaaS SFA to market. Most thought about SaaS as simply another delivery method. One company, however, saw SaaS as a revolutionary approach to software and knew in its bones that it would disrupt the whole industry.

    Today there’s only one company from that era still standing and you know who it is.

    Analytics is at a similar tipping point because it’s no longer the thing that the data scientists use to discover interesting things about the business. Analytics is a seminal technology that harnesses Big Iron to digest Big Data to give us the insights we need to compete.

    As Brynjolfsson and McAfee write in Race Against the Machine, it’s one of the things that enable us to play on the second half of the chessboard.  It’s what we have to do — because that’s inevitably where we are going.

    Published: 11 years ago


    Discussion

    • April 24th, 2013 at 1:41 pm    

      Interesting note: I was a user of SAS from the beginning, during its development stage, and concur with your assessment of that side of how things have developed. I now use R and see a similar evolution except, as you point out, everyone and his mother is trying to get into the data game.
      There are three points I would make:
      1: We tried to develop a AI statistician in the 70’s and have been thinking about it every since but the closest we seems to have got are the ‘plug-in and play’ systems such as SAS, S, R etc. Which are not Expert Systems but systems for experts.
      2: The problem is that you have to have a deep knowledge of your subject-matter [geology, ecology, marketing etc] in order to apply statistics well. I question whether most of the new ‘boys on the block’ have the dual capability: although they should be able to get it using a small team approach.
      3: Modern data analysis involves more than subject-knowledge and statistics, but these are the core. Certainly, programming beyond R is useful: I still advise people to learn ‘C, C++’ . If you know these, you have the basis of understanding most of the other necessary tools – which are now many and varied.

    • April 24th, 2013 at 11:57 am    

      You should use it in your site.Then you should consider ROI and what SaaS computing has done to it.think some emerging vendors are trying to do it in there .

    Speak Up

    You must be logged in to post a comment.