Jim,
You’re right, of course, the vast majority of companies don’t do even the most basic personalization. Given that, I think you’re point about real-time decisioning is intriguing. You said:
“I honestly think true real-time decisioning is overrated unless you are in select industries. I really do. I think you can do a lot with nightly models and reference the scores in conjunction with rules you have set that are grounded in behavioral analysis. Often that stuff probably doesn’t drastically change intra-day.”
I think that’s huge. I laid out a ton of different personalization options in some earlier posts and most of them didn’t require real-time decisioning. I also talked about matching your methods to your business and argued that cardinality is critical. Lots of products, lots of offers or lots of content choices all make for an environment where customer choice is unlikely to be stable and where effective personalization is almost certainly going to require real-time decisioning that goes beyond basic if-then rules. After all, if customer choices aren’t stable it’s highly likely that you’ll need to combine immediate behavior with past behavior to optimize intelligently. That's real-time decisioning.
Even where your decision-making is stable, cardinality makes a combination of nightly models and if-then rules tricky. You might not need a real-time decision engine to personalize, but you still need a way to match content/offer/product to model. If you’re pushing a lot of content, you’ll need some way to automate that matching process.
That doesn’t seem trivial to me and it isn’t something that testing tools make easy.
On the other hand, if you’re working in a traditional Website environment with a relatively static structure, limited amounts of new content, and manageable product/offer sets, you’re personalization opportunities are likely to be very different. Banks, insurance companies, brokerages, automakers (to some extent) and most other high-ticket manufacturers, life science companies, CPG companies – they all mostly fit that profile. I doubt that a real-time decisioning system is going to significantly out-perform your nightly models system in most of these industries.
Agree?
Let’s say you don’t need real-time decisioning but you do want personalization. Maybe this is a backward way to put it, but how do you figure out personalization opportunities? Because I see lots of companies struggle with this. They want personalization but have no real idea what it means in actual practice.
Now some people might think that this is a mistake; one of those hype cycle things where companies go off chasing capabilities because they think they ought to have them. But I don’t think it’s misguided to go looking for personalization opportunities. To me, it’s like looking for good bidding strategies if you’re going to an auction. You have every reason to suspect they are important and they probably exist even if you don’t immediately see the opportunity.
But I see lots of companies struggle mightily with this. Especially companies that don’t have that clear real-time decisioning need. It seems like the mindset necessary to identify personalization opportunities is challenging for lots of companies. I didn’t see this on your top five list of barriers (though maybe #4 is related). What do you think; is it as big a problem as I suspect?
If you were sitting down with a business to identify personalization opportunities on a Website, how would you go about it?
Finally, I was amused to see that the two Analytics Club threads crossed a bit. Kelly and I have been chatting about how frequently creative is a limiting factor on testing and I see it as #3 on your list of barriers to personalization. I’m not buying it any more here than there – and I think it represents a misconception foisted off on us by creative agencies; namely that most personalization opportunities involve a bunch of new graphic design choices. I think it’s even less true of personalization than classic testing. Have you been following that thread (if not, better go read-up!). What do you think?
Looking forward to more Krakken!
Comments