expertSEM

The next session I attended was in the Online Optimization track titled, “Behaviorgraphics: A New Approach to Online & Offline Analytics.” The speaker was Frederick Barber, COO of Directive Corporation. I thought this was an extremely interesting topic, but have to admit that much of it was over my head. When he started really getting into the nitty-gritty of analytics and data aggregations, my eyes glossed over, but I was still able to take some good notes.

I was trying to get a blog posted before starting another, so I was again a little late to this session. I walked in as Barber was talking about the process for behaviorgraphics. The following are the steps he mentioned:

1. Business Understanding
2. Data Preparation
3. Behaviorgraphic Modeling – predictive analytics; association analysis (use event trees)
4. Value Prediction — correlate behavioral segments to spend, purchase frequency and profitability; calculate (i.e., arbitrarily come up with) value for each; I see this much like placing a value on social media and leads; it may be arbitrary but it’s still meaningful

Essentially, what I got out of Barber’s presentation is that behaviorgraphics is all about making correlations. Much of it is assumptive, but can be extremely powerful if backed up by real data.

Text mining from blogs and product posting boards is one effective way to get this data. Measure count of posting events and categorize content into positive or negative and specific subject areas. He also mentioned this can be used in call centers: behavioral data can be summarized from customer service calls.

On to the Q & A section:

The first question was about people that clear cookies. The audience member asked how they deal with this and if there is a way around it. Barber stated that only about 90% of people actually clear them, so he has not really found it to be an issue.

Another attendee asked about using different computers (e.g., work, laptop, blackberry) and how they deal with this variation. Barber said they treat this information the same; you can assume the person using these different devices still has the same personality, so all data is useful.

Another person asked about combining grocery checkout data with online behavior. Barber said this is possible for not only grocery purchases, but retail and pharmacy as well. He said their are privacy restrictions and you cannot just go out and buy this list, but it is being done by some.

He went on to say that, for the most part, you can’t use credit card data. One exception is a major home improvement company (he didn’t say which, but you have a 50/50 chance of guessing which one). This company collects data using credit cards, but only when they get personal info at same time. They then track every single transaction when they see your card come through the store again. He didn’t say what they do with the data, but it’s an interesting fact. He also said that there’s a 40-45% unique match rate between last name and zip code when asked at retail purchase. Again, he didn’t state what this is used for, but it was interesting (if not somewhat frightening) statistics.

Another attendee (our very own Alex Porter, he insisted I give him credit!) asked about integration with Google Analytics and if it’s possible. Barber emphatically said yes. In fact, he stated, “GA is an open sandbox, you can build whatever you want.”

That’s it for the DMA09 sessions today. I’ll be back with more recaps tomorrow. There’s a social media session bright and early. I’ll try to make it but now, it’s happy hour so we shall see.