I'm running an experiment with Facebook Ads for my Tumblog
giantrobotlasers.com
giantrobotlasers.com
It would be interesting to see what would happen if you turned on comments for your blog. You also mention creating a "proper blog" which I think would bring something new to the mix as you build your brand. Right now your Tumblr theme is limited to what you want me to focus on at that moment, with little ability to get to your archives or see what you are all about (tumblr view) vs. a blog or RSS readership will know who\what they are subscribing to.
I think there is rationale to keep both, but you should try to build RSS audience at the blog vs. followers and tumbularity on Tumblr.
The reason to keep both is to keep things properly categorized. Pictures of cats -> Tumblr. Thoughts on a market -> blog.
For example, I could have "sign in" for accounts, using twitter or facebook connect. Then the "tweet this" link could be push-button, not the links to http://twitter.com/home?status=some_manual_posting_bs
RSS subscriptions that can also power email newsletters are certainly in the mix.
What text would you suggest? Obviously the title of the blog is pretty bad for the intent here.
Another tack, since it's a personal blog, is to target people based on similarities to you, rather than interests. "Founder" as a keyword targets 22,000 people in the U.S. "Robotics" targets 13,700. "Startups" + "startup" targets 4,700. If you add age-targeting on top of this, I bet your conversion rate from click-through to follow / RSS subscription would rise.
1) I think your target demographic, looking by the # of impressions. Try to go a little bit broader than just "tumblog". Try "tumblr", "blogging", "rss", .etc to reach a bigger audience
2) Your CPC is way too high. With FB it's best to go below their suggested bid, and gradually up your CPC amount until you start getting the desired impressions.
3) I'd gather a lot more data before making any kind of analysis on the data you're gathering.
4) Your CTR looks to be on the right track, although still probably not enough data to tell if it's stable yet. 0.1% is alright for a display ad on a social networking site. I'd still try split testing a couple different versions of your ad to see if you can up that. You always want to be split testing, constantly cutting the weaker performing ads, and making a variations of the better performing ads.
The comment there is really about others I've seen who, imho, aren't that talented, but get attention before they are good bloggers.