Hacker News new | past | comments | ask | show | jobs | submit login

Why didn't you wait to submit this until you actually had something to show, then? Could you even give a couple of tips other than "be good at targeting"?

Sorry, I'm not trying to be snarky, maybe I haven't had enough coffee today or something.




It's ok, I understand the frustration as I too am someone that has struggled with this and wished OP would have included said details.

Looking forward to his detailed write-up.


It's a pretty complex topic, so I'm still trying to figure out how to write it up coherently.

The shoemoney guide I linked is a decent intro to the topic.

The best one-liner advice I can give is A/B test a lot of images.

Also think about what keywords align with what you're trying to sell (obviously I can't really give very good generic advice on picking keywords!)


How long (time period) do you run your A/B tests?


I don't normally have a set period of time in mind, I just do it until it has enough views/clicks for the difference to look statistically significant.

I probably should be doing actual statistical significance testing, but I'm lazy, so just do it intuitively.


Should have at least put a picture of the ad that preforms well.


I'll be honest. The ad that performs best is a little cheesy, so I'm a little embarrassed to post it here where people actually know me :P

The image I used was a shrunken down version of this comic:

http://theoryofgeek.org/post/4050407389/the-optimal-brownie-...

It did 50% better than this one:

http://theoryofgeek.org/post/4291096322/date-date-revolution

It did 100% better than this one:

http://theoryofgeek.org/post/4205909563/the-social-networkin...

And about 200% better than when I used the profile pic image that I use on the Facebook page (from what I've heard on the grapevine company logos while great for branding tend to have very low CTRs):

http://www.facebook.com/theoryofgeek

(these datapoints aren't in the image I show on my blog; I tested them after I wrote most of that article and took that snapshot)

I should also point out that in CoderStack's ad campaigns cartoon images performed worse of all the images I tested, for CoderStack the best performing images were photographs of people. So there's no "one-size-fits-all" solution.


Cool, I guess one aspect of your best one here would be that it's more interesting the the equally easy to read logo. The other 2 cartoon images are probably hard to get at a glance at the small size you see the ad images at.

It's interesting because the limited campaign I have dealt with before used a logo and tagline type display, probably not the best.





Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: