Data Mining: Gluttony or Grail?
Wednesday, May 21, 2008
I recently went with Maddie "Get the data, then throw it out!" Grant to an ASAE Technology Idea Swap on data mining. Several folks from inside ASAE were there showing off their new predictive modeling software package, Clementine. Predictive analytics, in my mind, is like the holy grail of data mining—it becomes an obsession for those who pursue it, but does it really change anything?
As a marketer, the data geek side of me is just dying for a magical piece of software to make decisions for me. Who should I target? What should I say to them? What’s the magic formula for getting someone to join/register/buy?
But true “Freakonomics” moments—where the singular pursuit of data analysis reveals a truth no one ever could have hypothesized—are rare and overrated. By overindulging in a gluttony of data, you lose track of cause and effect and weaken your findings. For me, the real value of data is the ability to test ideas from real people who are bringing all of their experience and real-life interactions to the table. In other words, learning directly from our co-workers, partners and members, and then double checking what we’re learning against our data.
It’s no accident that Google has built human bias into their algorithm, or that human-powered search like Mahalo is all the rage.
At the ASAE Tech Conference in January, David Gammel and Wes Trochlil presented on putting the intelligence back into business intelligence. I was really impressed with their insight, and so I’m tagging them both on this post. I'm also tagging Mads who was at the swap;-) I hope you guys will share your considerable knowledge on the topic of data mining.
So the question is, does it make sense for a membership organization to spend time and resources on predictive modeling, or is it just another layer separating us from our members?
6 comments:
I'm presenting on using BI in membership and marketing next week at the ASAE Marketing and Membership Conference. One of my overarching points is that an organization must have a willingness to act on the information it has. So even predictive analytics are useless if the organization doesn't leverage the data to change how they do things.
This is still very much worth it in the long run but I agree with you the the 'hallelujah' data results aren't going to happen all that often. And, as you and Wes have discussed, it all doesn't matter if you aren't willing to act upon what the data show you.
I know I get wrapped up in the nice-to-know-cool-flashy-stuff versus the need-to-knows and you can use predictive analysis until your blue in the face and not get any actionables out of it. It has to be about changing behavior. And then measuring it. And then measuring it again.
We’ve just started on our predictive modeling journey. Love to hear all the view points. This “stuff” is relatively new to me, so I’m looking forward to learning from everyone along the way.
Part of me would really like to be proven wrong here. If I can offer any advice to you and your team, it would be to 1) identify the decision-points you might be able to influence, 2) convince the people making those decisions to buy-in by bringing them into the process early and often, and 3) tap the knowledge of your colleagues to differentiate between cause and effect before you add a new variable to the analysis.
Also, check out Maddie's thoughts on the same idea swap in the "Links to this post" below.
