BeyeBLOGS | BeyeBLOGS Home | Get Your Own Blog

« January 2007 | Main | May 2007 »

March 17, 2007

Two dimensions are better that one...

I've for a long time made the point that we're living in a 1-dimensional world when it comes to displaying data. Newspapers and magazines displays tables, barcharts, and piecharts - but nothing more. In the back end of the Economist you might see a scatterplot, but in the WSJ, Financial Times, Washington Post, you name it, we'd never see anything more interesting than the most simple depiction of data.

So what a pleasure to see NYTimes being gutsy enough to not only try but also keep up an interactive display of the stock market. If you haven't seen already - definitly do so!

You will find it in the business section of the NYTimes

Share: del.icio.us Digg Furl ma.gnolia Netscape Newsvine reddit StumbleUpon Yahoo MyWeb  

Posted by Christopher Ahlberg at 10:45 AM | Comments (17)

March 1, 2007

Long Tail of Analysis

The analysis of data is obviously a very large space. We do analysis absolutely all the time, e.g. figuring how a restaurant bill adds up or how my departmental budget is working out. Small tasks and big tasks. Certain analytical tasks end up getting lots of attention and even gets dedicated analytic applications - famous examples include software for optimization of retail space, credit scoring, and customer segmentation.

Recently the book Long Tail has gotten a lot of attention. I can clearly recommend it, great book and also short and elegant(!). The basic premise of the book is that in the past limited resources like shelf space, movie screens and tv channels, expensive resources like shipping and storage, and difficult tasks like finding willing buyers, and sellers, etc. have forced businesses to focus only on the top sellers in all categories or the vital 20% of the 80/20 rule.. The rise of cable television, the internet, , the browser, Fedex new technology for data storage, user access, networking, etc. - allowed innovators like Amazon, Netflix, and Ebay, to open up new markets to the 80% of customers and niches not served or the “long tail”. For example, an obscure movie from 1950 suddenly is available and can be a money maker - this is part of the attraction of market places such as Amazon and Netflix. This ability to not only serve the "long tail" of customer demand but also make it a very good business proposition is the premise of the Long tail.

When I read the book I clearly thought of how it related to analytics but really only thought about what kind of analytics we at Spotfire could do on the long tail itself. Then last week a long standing colleague of mine, Tim Loser, emailed me and suggested something quite clever. Basically he made the point that the actual long tail of analysis is quite interesting in itself. And he is right on!

Think about it - as I mentioned before certain kinds of analytics has gotten its own software - analytic application and that's great. I think IDC calls out some 600 companies doing analytic applications - but I also think it's fair to say that there are tens of thousands if not hundreds of thousands of variations of analytic applications. Just think of all the different Excel spreadsheets out there capturing various kinds of analytics - in your company alone I'm sure there are hundreds or thousands.

The obvious question is how has this [very] long tail of analysis been served to date?

Interesting observations can be done here: some times it's served by IT departments building applications serving business users. But only a few can obviously be served that way by traditional means. It's expensive, it's heavy - both in development and maintenance. I meet plenty of CIOs telling me “I’d love to provide better analytics to my business users but I’m drowning in maintenance of what I’ve already got”.

Another approach is where the tail is served by in-house analytics groups - they exist in many large companies. A similar approach is external consulting firms - everything from small one-man-bands to McKinsey helping out companies with analytically oriented work - delivering everything from glitzy Powerpoint presentations to highly configured Excel models. In fact this business is a very large business sector - and can be very profitable for the consultants, even though it's man power intensive, and not exactly scalable (or economic?) for the customer.

Now - if we assume that the long tail of books, films, music, etc. became available to customers with the arrival of new technologies - what happens with analytics with the arrival of new technologies? Shouldn’t we be able to fundamentally change the world of analytics here?

The first step is what we’d like to think we’re doing at Spotfire, by providing a platform that is extremely easy to use but also very to configure to a particular task, we can have one platform serving many many users across many many business problems. Good for end users, good for the CIO and the IT staff. Business analysis groups like [should like] it as it allows them to spread not just conclusions but also share methodology and approaches.


The next step is the really interesting. What are the economics of analytics in a digital world? One obvious one would be the “iTunes of analytics”. We have seen ManyEyes and Swivel (I vote for ManyEyes) as interesting examples of visualization of data on the web. They certainly have a lot of content, but the question is if they really help an analyst doing analysis? I think they do allow somebody to share an analysis and point to it – but they are still a ways to go to provide this in a form where you can find an analysis that can help you solve a particular problem, and also allows you to apply it to your own data, etc. A really attractive version of this certainly could be helpful to a lot of people – and truly be the iTunes of analytics.

Another obvious one in a world of digital analytics is how the role of the consultant changes. Instead of giving glitzy PowerPoint presentations with canned conclutions, taking questions, billing more hours till the next presentation, repeat, etc. Innovative consultants should provide analytic applications where customers can ask and answer their own questions and that can be reapplied time and time again. You may say that that would be too hard for them (programming, etc.) – but I don’t think that’s a valid excuse anymore – easy to use platform now exist. You may also say that the economics works against analytic applications – consulting firms make their money by doing the same PowerPoint presentations over and over again. I say “welcome to an outsourced and down staffed world”! The consultant who not only wants to survive but also move up-market will move from PowerPoint to providing reusable analytics. A consultant who wants to open up the long tail analytics market will provide reusable analytics.

What else changes in a world of digital analytics? What does IT need to do differently, even with a great platform, to serve the tail? Should a company trying to gain a competitive advantage (Competing on analytics) focus on the head or the tail – easy to say the head, but is it in the tail the difference really is made? Is there a new class of knowledge worker who’s job it is to orchestrate how a company manages its tail of analytics? Is there a role for academia in this? What’s the supply chain of analytics that allows analytics and smarts, not just data, to be shared across companies – from Walmart to P&G, from P & G to the suppliers and back?

Share: del.icio.us Digg Furl ma.gnolia Netscape Newsvine reddit StumbleUpon Yahoo MyWeb  

Posted by Christopher Ahlberg at 11:00 AM | Comments (24)