Just as some people define they’re identity by what they buy, some people define themselves by the tools that they use.

There’s a certain cache about using the ACH. Or being an OCP. Or knowing enough to choose select instead of forward regression. Or the use of Bayesian methods. Coremetrics against Omniture. Google Analytics over Webtrends. R over SPSS. Graffle over Visio. And so on.

There’s a large degree of tool centricity in three communities: web analytics, data mining, and marketing science. The irrational judgements about people in each of those communities, based on the degree of sophistication of tools, is dangerous. Worse – it’s detrimental.

It’s detrimental because it narrows your view.

For one, different tools are right for different lines of inquiry. Sometimes the solution is to use a form of Bayes. Sometimes the solution is to use a genetic algorithm. Sometimes the solution, to be driven down into a scalable algorithm, has to be possible in SPSS or in SAS. Sometimes simpler is better. Sometimes complex problems need to be addressed with complexity.

Having been through all three communities, I can say that Marketing Scientists are by far more statistically sophisticated than data miners. By several orders of magnitude. Marketing Scientists at this last INFORMS conference were working on problems, using complex methods, in efforts to clean up rounding errors in some of the most well researched areas, long hollowed out by the last cohort (auctions and competitive game theory come to mind!).

Data miners, on the other hand, use methods that are several orders of magnitude more complex than web analysts.

And yet, certain web analysts, examining log file data with their own commercial tools, using their methods, have far more understanding of how a complex system like a web site performs, than certain marketing scientists.

I’m not advocating aloofness.

I abhor the aloofness that goes with the generalist mindset. If you know just a little about anything, you really don’t know much about something. You need to have a very strong base to be able to come out and engage others. You have to have something to contribute to other disciplines.

In 2011, I’m really going to try to think outside of toolsets. I won’t let tools define who I am.

Tricky. We’ll see how that goes.

3 thoughts on “Definition by Tools

  1. Happy new year! 🙂

    Your post touches on two points: first, when I hear people complaining web analytics is hard I can only assume they have a very narrow experience and were never exposed to much more complex environments and challenges.

    Secondly, it’s been said, and I have quantified through my work on the Online Analytics Maturity Model the fact the toolset doesn’t have much to do with how successful you are at web analytics.

    Let’s make a resolution for 2011: 1) go beyond the tool and 2) tackle even the most difficult situations with a true analyst spirit: deconstruct even the most complex situations to have a better understanding of them and optimize them… be it our own web analyst role or any online activity.


  2. Christopher and Stephane:

    I strongly commend your efforts!

    Perhaps problem solving, that skill being inherently hard to interview for, could be better understood by looking at those variables which correlate with it.

    Studies have shown increased language skills are just one of these variables, the ability to frame a problem in a foreign language appears to offer new vantage points.

    Has a recruiter ever asked a candidate about the different languages they are familiar with?

    In our realm I would add programming languages and econometrics/applied statistics as a predictor of success. Not because you are necessarily going to perform advanced calculations every single day, that would be too awesome of a job to exist.

    Rather, those fields require constant improvement and curiosity. Programming languages move with new releases, econometrics/applied statistics both require constant reading of articles to stay up to date on the latest improvements.

    Firms looking for candidates often look for the absence of something to reject a candidate. Know Omniture? No? Rejected.

    When they should be looking at what the candidate does have:
    * Indicators of problem solving skills
    * Self-motivation to improve their skills
    * Innate curiosity about their environment

    Michael D. Healy

  3. @Stephane Thanks, agreed. Let’s do our best. (It won’t be easy breaking out of old habits).

    @Michael True. Recruiters are trained to look for a list of terms. The best recruiters use these terms as filters, then talk to people to verify. Unfortunate yes, common yes.

    There are times when a firm is looking for a specialist to solve a problem. In those instances, knowing Omniture is essential to solve that particular problem.

    There are often times when a firm is looking for a generalist to operate a very defined set of instructions. Only that the firm hasn’t actually really defined that set. So a large laundry list is produced. That laundry list is typically tool specific.

    Calling myself out, I do list a ‘one of four technologies’ section on our job postings at Syncapse. I call for SPSS, SAS, JavaScript or MYSQL. And there’s a very good reason for requiring one of those four technologies. I demand knowledge of the scientific method – which is another tool.

    It’s predictive of having a concrete base.

    Most jobs mandate experience with a given toolset.

    People who can operate reliably are rare. People who can optimize their own operation are even rarer.

    So, there’s a bit of a meta-meta-meta aspect going on.

    To the language point – to a computer scientist – all languages are just syntax. It doesn’t matter. To somebody who has been trained to think very narrowly or taught only one language – well. That’s a deeper problem.

Comments are closed.