I've used A/B testing for e-mail newsletters and experimented with it on landing pages. Basically everything can be tested to see which version works better, gets more clicks or converts better.
News site The Huffington Post has been using A/B testing to test headlines for a while now. Their huge traffic gets split for five minutes to decide which headlines has a better click-through rate (CTR).
For lower traffic websites, this will take a bit longer. But even if it takes a week, that best headline will stay on your site and show up in searches, ensuring better results.
Another possibility is to use your newsletter or social media accounts to test which article headlines are more powerful.
But newsletter traffic probably converts a lot better compared to general site traffic, so you might be losing some opportunities there. I would keep headline testing on your site and use e-mail newsletters for better purposes.
It is not that hard to write headlines that attract the attention. But coming up with good titles it is still really hard.
The challenge however is to transform your article, which better be interesting, into a good title. That way people who clicked through from wherever will stay to actually read your article. And they won't feel cheated afterwards!
A/B testing headlines, a good idea? Drop your ideas in the comments!