I'd like to add a few things to Thursday's post about "using click-through data as a segmentation attribute for targeting".
Not many people are aware of how email tracking is done. So being subtle is the keyword when you use click-through data as a segmentation attribute for targeting. Be careful not to give the recipient that "big brother is watching you"-feeling because you will scare him of for sure!
Instead you could refer to the fact that you sent them an email "last week" and refer to the fact that they might have read it.
It's the same when you would do telemarketing follow up of your email campaign. If you were to call someone that had clicked-through on an offer, you shouldn't say "we saw that you clicked on our offer", but you should just refer to the fact that you sent them an email on the subject of XYZ and ask them if they had received it/read it.
At what point do people get upset? That really depends on the type of message and on your target audience. Concerning the type of message I would say that resending the same message to those recipients that didn't open the original email (Keep in mind to always use a new subject line when you do this!), is less of an issue than resending the same offer to those that already opened the email but didn't click through.
My personal feeling is that if someone didn't click on the offer after the 2nd or 3rd time, they're just not interested and you should try another offer. One of the tricks you could use, is to include an alternate offer in the follow up email next to the primary offer.
About the exact frequency, the best advice I can give anyone is to test it and monitor your unsubscribe rates very closely. Personally I think that you should leave 1-2 weeks in between each follow up email that you send to the same recipients.