Ad Split Testing Method for Content Campaigns
September 21, 2009
Content campaigns should be separate from Search campaigns, granted…
The methodology I am suggesting is based on a few underlying premises:
- There is some difficulty gaining “traction” in the content network
- The Ad testing “life cycle” must be longer than search ads
- True Ad testing occurs only when ads are created at same time!

photo by: Rat Goddess
The main differences in my content testing methods, compared to search are as follows:
- Begin with 5 ads, versus 2-3
- Spend more time creating content ads; the testing cycle will be comparatively much longer
- I commonly delete losing ads, ie. whittling down from 5 ads to maybe 1 or 2, before I'm ready to begin aknew with maybe 3-5 brand new ads, with no stats.
- I less frequently will "start over", ie. erase all ad stats, and begin the horse race again, and when I do, I will be careful to do this on a Monday or Tuesday, avoiding the weekend approval lag time.
You may decide to only work thru several adgroups at a time. You may prefer to create totally new ad concepts or concentrate on formatting variations, you decide.





5 comments:
Hi Dan,
Clever first three steps - they make a ton of sense.
Question, though - why delete the winning ad, rather than just pausing it for the duration of the new test? Don't you want that history working on your behalf?
Tally-ho,
Howie
Dan,
Interesting concept, but can you provide more information on the background of why you suggest this strategy; for example, i've always held in the content network you should never delete ads, only disable them; and why is using 'optimize' vs. 'rotate' beneficial in this case?
I've revised this article, opting for only rotating ads, not optimizing, bad idea.
In conclusion, content ads optimization begins with "whittling" down losing ads to 1 or 2 ads, and less frequently "beggining a new horse race", with brand new ads.
it really very interesting. I was always facing problem with my ads in content networks but now this information will help me to test my ads in content network. it would be more helpful if you post something in more detail. very nice post.
Would you have any test results to share over on A/B Tests dot com? We'd be interested to see some data from your testing methodology.
Post a Comment