First Rule of Behavorial Marketing: Do Nothing Creepy

An excellent commentary from Kara Trivunovic, senior director of strategic services at StrongMail

Systems, highlights the need to not be "creepy" when using behavioral tracking (what a prospect clicks on or downloads) to decide what follow-up offers or info to send to them.

Among other things, Kara recommends being careful about being too literal around interpeting clickstream data at a very granular level, such as blindly spamming a prospect with information on a line of shirts just because they looked at one shirt. However, he says, "if you are able to piece together some inference around a categorical interest, you have more footing and can then send an offer supporting that category."

Has anyone out there build such inferences for B2B sales (especially in the IT space)? If so, what "chain of content" has worked for you, and what tools did you use to build the inferences?

Comments

[Seamus Walsh]

Bob,

We use a series of activities that puts visitors through a maze of content across 5 websites to determine time spent on site and participation of various polls, survey's and situation assessments.
We give visitors enough to do to keep them busy and coming back. Our poll functionality allows only one vote per IP address.
We sample that data across time and delete logs every 90 days to reduce our creepiness.