I can pretty much measure how tough a week was by how often I had to dress up. So last week (with client meetings in NY on Monday/Tuesday, OMMA in SF on Wednesday, the Red Door Speaker Series in San Diego on Thursday, and client meetings in Palo Alto on Friday) pretty much scored a perfect five. My sneakers got a much need rest but my voice is a bit scratchy!
OMMA and the Red Door Speaker Series sparked some ideas that I wanted to share before I embark on a more technical series on use-case analysis (and maybe a few more segmentation posts).
The hook for the OMMA panel was the “Metrics Meltdown” – why analytics is failing. And, like most panels, the discussion probably revealed more about the panelists than the topic!You had the consulting guys (like me) saying that problem was largely about people. The tools guys (Brian Kelly from Quantivo) blaming old-generation tools. And the enterprise guys (Adam Greco from Salesforce) talking about enterprise politics and culture.
Each of these is a legitimate gripe of course. And each is reflective of our own parochial view. My business is all about finding and getting good people. Brian’s is about building better tools. Adam’s is about getting things done in a specific corporate climate. But I left wondering if we all weren’t a little off the mark. The more I think about it, the more I’m skeptical of the original question’s premise.
Is online analytics really much different than any other corporate endeavor?
My background is in traditional IT (programming) and traditional Business Intelligence. Based on my experience then and what I hear now, I’d bet my whole small fortune that if I went to a team of BI gurus and asked them the state of enterprise reporting they’d say it’s a mess - enough to make you despair.
When I hear about corporate IT, I rarely hear that it’s in good shape. My clients certainly seem to dread IT rather more than measurement.
So is web analytics really that much (or any) worse than anything else? Get a group of experts in any business area and they’ll tell you that most companies are very poor at their particular discipline.
They wouldn’t be wrong either.
If you can commoditize a skill or discipline, you can expect many companies to be good at it. In IT, functions like PC Setup or Network Configuration have become almost commodities. Advances in software and hardware have made these tasks much easier, the high-level of standardization of the tasks across companies makes learning highly transferable, a mature understanding of the disciplines have led to solid best practices, and high-levels of training and certification have combined to make these tasks that companies should expect to be competent at.
Other IT skills have resisted commoditization. Programming, database administrators, data architects, and modelers have remained, and show every likelihood of always remaining, highly customized non-commodity disciplines.
Analytics, whether in the form of traditional BI or in the online world, is (thankfully) not a commodity item. It cannot be automated. It is not standardizable (in useful form) across different organizations. It cannot be captured in any tool.
The idea that good analytics can be baked into a tool or commoditized is no more sensible or likely than a similar view about making TV Commercials. Just because both benefit from and rely on technologies doesn’t mean that the technology can or will encapsulate the value. Each requires a significant technical skill that must be combined with understanding of a specific business and their audience. It is the art of this combination that drives value – and there is simply no prospect of this art becoming a commodity.
In thirty years of significant BI work and investment, there has been absolutely NO movement toward commoditization. The hope that such movement will somehow occur in web analytics is forlorn.
What are the implications of accepting that online measurement and analytics isn’t a commodity?
Most companies do only a few non-commodity things well - usually the thing(s) that are core to their business. That thing isn’t usually analytics.
So in circling back to the original question, it seems to me much less than obvious that there is any such thing as a metrics meltdown. We are not obviously worse off than many, many other non-commodity disciplines. To me, the question shouldn’t be: “Why is our discipline so messed up?” The real question is “How can any one company become an exception to the general rule of doing any particular non-commodity skill poorly?”
Everyone talks about PPT – People, Process, and Technology as the keys to success. It’s stock stuff and its stock because it’s true.
But I’m going to hearken back to my original answer and focus on people.It’s not that process and technology aren’t important. Good process and good technology are essential – especially at the large enterprise level. They’re essential because you cannot count on having good people. But good process and technology can never do more than assure some basic level of competence. They are not enough to achieve excellence.
If you are sports fan, you know that in football (whether collegiate or professional) having a great coach is essential to continued excellence. While football requires individual excellence, that excellence needs to be controlled, directed, and recruited by a strong guiding hand. I think analytics is similar. If you want to excel at analytics, you need to treat hiring an analytics manager as if you were hiring the head coach of an NFL team.
You find the best – and you pay them what it takes to bring them in.
In our field, hiring a star doesn’t mean hiring some guy with ten years experience and nothing to show for it. Or assembling a team from your existing departmental marketing resources and sending them off to Omniture training. That’s no way to be really good something. If your idea of being good at web analytics is to hire people and send them off to Omniture certification, you are going to fail.
I see many major enterprises spend loads of money building measurement teams that never accomplish anything interesting. You have to spend at the top first. That means hiring Adam Greco or Joe Megibow – it means shelling out some serious money for them and then for their team.
Is it worth it? Yes – guys like that are night-and-day better than your average analytics manager recruited from your existing marketing department and they are (I believe) worth far more than they are paid. You want great, hire a star with all the investment that implies. You want excellent? Hire someone like Semphonic and let us really loose (thankfully I know there are only a few stars to go around). You want be at least competent - get some good, basic processes and technology in place. Do less than that and you’ll be looking up at mediocrity.
This isn’t because of a metrics meltdown. We are not, I think, a special problem child. It’s impossible to be really good at everything. It’s hard to be really good at any ONE thing. If online measurement and analytics is really important to you; if you have a commitment to excellence not competence; the likeliest path to getting there is not much different than it is for countless other disciplines.
Hire the best; let them take care of the rest.
Comments