In summary – yes and no. You want to try both ad types at the same time and see which one does better over a 30 day or longer trial period. Then eliminate the underperformer.
I’m using this post as a very extended annotation in google analytics because for some reason it only allows 160 characters. What is this, 1995?
Anyway yesterday (June 4 2013) Google asked me to allow images as well as text ads on my highest paying slot, a 336×280 at the top of a certain website about buckets. They came to me promising up to 30% increase in weekly earnings, which is the same as a 30% increase in monthly earnings actually, if you do the math.
The unit was making about 10$ a day. For comparison, rent costs about 13$ a day, so this is a fairly important ad.
This is the 3rd or so time they’ve come to me begging me to allow those image ads. I’ve always clicked no but I figured – why the heck not – and clicked yes this time.
Day 1 of Rich Media Adsense Ads
The next day was depressing. Ad revenue for the entire site actually dropped by half.
I looked to the internet for some advice – and wouldn’t you know it I found some.
Apparently image ads used to have not such a great reputation back a few years ago, but more recent experience seems to indicate that they DO improve earnings. They may actually attract less clicks but since a single image ad may be buying out 4 smaller text ads, the payout per click can be much greater.
But the people who know better than me advised, each site is different. Which is true. Some sites get better responses to images and some text. Google’s algorithm is only as good as the average.
There’s a simple solution – test both at the same time for an extended period of time to get some hard numbers.
A/B Split Testing
I’m using 2 easy to use, free tools that any adsense site should already have set up.
- Adsense Ad Channels
- The ad rotation feature in my wordpress Ad Injection plugin.
Instead of just serving one ad in this spot, it’ll be a randomized test between ad A and ad B.
- Ad unit A: text ads
- Ad unit B: text or image ads
These will run either-or for 4 to 6 weeks, and after the end of the period we’ll see which one pays off better.
Adsense Ad Channels
I have 2 ad units that are set to serve 50/50 to each visitor. My new one is text only (same as before) and my “upgraded” old unit is now text and image ads. Both ads are the same in all other ways, including placement and size.
I’ve set up 3 ad channels. One for each of these ad units, and a third that adds the 2 together.
Based on the reports from other people who have taken Google’s suggestion to allow media ads, I expect new relevant advertisers with image ads to trickle in over the next month. This test will allow me to get those hard numbers that I can confidently go forward with. I’ll be able to calculate a percentage and see how close I came to the alleged 30% claimed. Right now I’m at -50%, so we’ve got a ways to go.
However I expect that the rich media slot will earn more than the text slot by the end of the 30 day test. But there’s no way to know without doing the test right.
Results will be posted in a separate post. If I remember.
I have a few concerns that could muddy the data.
This new ad unit has no experience, it’s like a level 1 warrior on final fantasy. The older ad unit has been running for 6-8 months. I don’t know if the new ad unit will perform at the same level quickly enough to be a good data point. I don’t know enough about how ad units work within the Adsense system to know how quickly my level 1 ad will reach level 6-8. It could be a day, it could be another 6 months. There’s no way to know unless I can gather data from a much longer period of time.
This is the kind of problem that would eat at Taran. It’s at this point I get lazy and just let things slide. I may not have the patience for science.
Notes to Self:
Throw out data from June 5 and before. June 6 2013 and forward is good data.