Talking Personalization
[Jim Hazen and I have been talking about Personalization and the appropriate technology stack since his days at SAS before he joined Semphonic/EY. Back then, we talked about doing a white paper together on the best approach to creating a personalization stack. Since that never quite came off, we’ve decided to tackle the topic as a dialog in my blog. So over the next 1-2 weeks we’re just going to exchange posts and thoughts on the topic.]
Jim,
First, let me say how excited I am to be talking this over – it’s been a long time coming!
If I remember right, we first got into this topic with a conversation where you mentioned how surprising it was that most enterprises have tackled personalization with testing tools – and how limited those tools were in terms of the ability to blend customer data with real time actions.
That ability to blend historical data with current actions does seem to me to be at the heart of what you normally want to achieve with digital personalization (though if you go through the types of Personalization tactics I described earlier, a goodly number work exclusively on one side or the other and don’t require blending). Is that the right way to think about and judge the spectrum of personalization technologies that are out there – from testing tools to black-box approaches to solutions that harness multiple technologies in a hybrid approach? Or would it make more sense to figure out what tactics you want to implement and then pick tools natural to those tactics? Or will it amount to the same thing since almost everyone will want to employ at least some tactics that require blending customer data with real-time, occurring just now actions?
It seems to me that your skepticism about testing tools in particular was a pretty natural perspective for anyone working at a company that specializes in delivering powerful statistical analysis tools. Most testing tools have zero statistical modeling capabilities and, truth to tell, even their rule-engines are kind of primitive. Has time on the consulting end changed that perspective at all – or has it reinforced it?
Comments