What I'm talking about is really an interaction analysis for specific segmentation groups. Then using that data in ralation to varied send-times in order to spot behavioir patterns.
For example: We send a Channel newsletter out on a weekly basis every Wednesday afternoon at 2pm ET. It contains promos, contests, original content our partners can use to sell, training webinars, etc. I'd like to create a report that tells me when (day/time) people are opening this message and when they are clicking on links in this message. Then vary the send time to 10am on the same day, then roll it back a day, etc. Take all of the snapshot results and compare them and you'll start to spot trends. the same thing can be done with any other subset of your database.
The physical chart would be laid out like a calendar, days are columns, rows are hours in the day. Stat with the day of the send and look 5 days forward. Each box contains the percent of opens or clicks (or whatever you're measuring) in the cell. Shade the cell from red to yellow to green depending on the percentage of interation. A quick visual comparison of your snapshots will lead to some emerging trends for that segmentation group.
The ultimate goal is to determine the optimal time to send messagin to a specific group. A more sophisticated approach would be individual-level interaction optimization the way Silverpop has implemented, but for now I'd settle for this.