I just nuked the Google Adsense stuff from this weblog. I had added it to see how it did with analyzing the content of various pages (along the lines of the bags-of-keywords and content-indexing memes with respect to learning objects and metadata).
The experiment was actually quite interesting and informative. Conclusions:
- It is definitely possible to effectively match content (ads) with other content (weblog posts) based on the bits within that content (and not on keywords or taxonomies applied to the outside of said content).
- The matches between content and ads are often extremely accurate, but lag by a couple of days as Google updates indexes and caches. New posts tend to have a quasi-generic ad for a day or so, and then a "real" ad kicks in
- Readers of this weblog (perhaps/not weblogs in general?) are rather task-oriented folks. They don't get distracted by ads - they are apparently here to read stuff, and don't tend to click on ads
- As a result of the previous point, Adsense isn't a viable revenue source. It wasn't intended to be - I would have donated any revenue to the Breast Cancer Foundation - but over the past several months, Adsense tracked a grand total of less than $10US. They don't pay until you hit $100US, so I would have to keep ads running for almost 2 years to see a dime from Adsense. The Adsense EULA prevents me from publishing specific details on success rates (CPM, etc..) but they were substantially less than stellar. ;-)
By the way, the conclusion of this experiment was triggered by the IT Conversations: The Future of Online Advertising (at Gnomedex 4.0) session. They got talking about "monetizing content from weblogs" and adding ads to rss feeds. It left such a bad taste in my mouth that I couldn't continue with Adsense. The talk was quite interesting (with folks from Google, Amazon, etc... talking about what they do and where they are going), but a wee bit scary.