With two very intense days of 5-track sessions going at NCDM 08, it was difficult to pay attention to everything that was going on. Still, by picking up handouts from sessions I did not go to, I could get a sense of what the hot topics were this year:
1. Web Plays Data Friendly - Almost every session had a web component in it, the BI folks are getting it done. The data is coming off the web / out of web analytics and into the data warehouse so everybody knows what is really going on.
Much of this work is being done by the big service bureaus, not the companies themselves, as far as I could tell. Web Intelligence, baby.
2. Media Mix Models – which typically show online Display advertising as a poor choice for allocating marketing budget to when you are also running offline media; the yield is quite poor versus almost every other media. Search, as you might expect, Rocks on Productivity, though TV still rules for Productivity and moving the needle.
The implication here is you’d be much better off running TV to generate / amplify Search behavior than running Display to do the same. Offline for Awareness, Online for Intent / Desire.
3. Contact Optimization – I’ve written before about what ultimately happens when you don’t have a Contact Strategy. At some point, BI will measure the bottom line impact of every division in the company pounding customers with the division’s own contact strategy (Hint: you are driving your customers crazy). Then, move all customer contact to a centralized model which controls contact by source of new customer and value generated, measured through controlled testing.
This movement should not be surprising, given the whole “customer in control” and “social” movements which onliners give so much lip service to but never take action on. Well, not never, but rarely.
4. Measuring Engagement – yes, Engagement. OFFLINE, as well as online. The overwhelming message was this: it does not make any sense, and actually costs you money, to keep pounding your customers with any kind of communication when they don’t respond / interact. You can measure dis-Engagement, and when you see it, you should stop communicating – online or offline.
And these folks proved it, over and over, with real math. See related #3 above. If you’d like to see a detailed e-mail example of this concept, check out this case study.
And of course: Models and more models. Like Hierarchical Bayesian and Disaggregate Discrete Choice.
Boy, I love it when you talk like that.
