I do think one should always have an offer of sone kind. Even if its an email subscription to a blog. That way you can develop a redership for possible future offerrings. Specially useful if you see startup in the near future.
This is measuring something different though. I could have a bad article that heavily pushes an offer and a very good article that doesn't.
The time on page can give you a relative statistic to compare your articles and develop a more engaged readership over the long term, not just optimize to try and sell them something right now.
An example would be something like patio11's posts, some of them are pushing for a conversion but I don't think that each would be best measured in terms of a conversion made right now.
There are three factors you have to keep in mind when writing content: Message (the content), Media (how you distribute it (a blog)), Market (who reads it). The better way of adjusting for those three (optimizing) is to measure conversions on the Message. Otherwise, you are tracking fluff. Of course, my opinion based on experience.
