Best Way to do Ads Split-Testing?

[bigdealioo]

Well, basically I took the advice of making a lot of ads to heart and wrote like 12 ads. The problem is that each of them got very few impressions and I think it's gonna take a year before each will get a reasonable estimate of CTR this way.

Should I cut down the number of ads being split-tested to say.. 3? Then after that, pick the best and add 2 more new variations? And so on.. rather than testing all "thinkable" variations all at once?

Maybe, at first pick 3 very different approach ads. Then get more and more "detailed"... each picking the best performing ad and trying progressivelly more and more micro-level variations on it?

WHat's the optimal way?

[lethal0r]

i just pick 2, most of my keywords dont have many impressions so it would be futile to do 12 ads at once. i run 2 ads up to maybe 200 impressions or so and see how they perform. then delete the worst one, make a note of what it was and try try another.

[trannack]

Adword sspecifically advised me - on more than one occasion - that 3 ads was optimal. I ran the three for a couple of months, then weeded out.

[eWhisper]

FYI - related post:
[webmasterworld.com...]