A/B split tests while starting a new campaign

[crocwatcher]

Any thoughts on the timing of running A/B split tests with a new campaign. Should I be concerned about one of my split test pages being too short (long form vs. short form landing pages) and therefore lacking enough content to get good quality scores, and hurting my cost per click if the campaign is new and prices are still dropping on a daily basis? maybe wait and do split testing after cost per click has stabilized? Any thoughts would be appreciated!

[Trace]

I prefer having stability on the Quality Score and click-through rate before I start running an A/B test. Unless you have a great CTR and a high Quality Score right off the bat, I would take some time and focus on the ad side of things first. Once you have found an ad that is engaging you will have a better feel for what the copy on your landing page should be.

[AussieWebmaster]

I always go for two ads to start.... more opportunities for keywords to align with certain ads

[crocwatcher]

hmmm.... two replies, two different answers, LOL! Guess I'll ponder it some.

[undercoverseo]

I've learned to start with 2-4 ads in an ad group with varying types of ad copy (think extreme ends of the spectrum; conservative vs. flashy ad copy...different calls to action). We let it run until we get a good sample size of impressions (above ~400) before we start weeding out the ineffective ones and tweaking further.

This method helps determine right off the bat which type of ads work for us.

[chris1]

I think what he is talking about here is split testing landing pages, not ads. If its ads, I would recommend testing right away.

Landing pages however, I would set up 2 identical campaigns with only the destination urls being different. As long as you keep the bids the same, you should see close to 50/50 split traffic. Keeping seperate campaigns allows you to track quality score differences (in addition to conversion rates obviously) for the seperate landing pages. of course, this is not the optimal way to manage a campaign so I would switch back to a single campaign as soon as you had significant data to draw conclusions from the test.

[crocwatcher]

You are correct Chris, I am referring to landing pages, not ads. So you would not use a/b split tests inside of google's web site optimizer then? Have you had any bad experience with using google's tool for this? I was thinking that by using the google tool, you would avoid any slap action for duplicate content from google, since they know you are testing splits. Thks.

[undercoverseo]

Originally Posted by chris1

I think what he is talking about here is split testing landing pages, not ads.

Ah, my bad. Continue on with the regularly scheduled program...

[crocwatcher]

no problemo!

[khalidh]

Originally Posted by crocwatcher

You are correct Chris, I am referring to landing pages, not ads. So you would not use a/b split tests inside of google's web site optimizer then? Have you had any bad experience with using google's tool for this? I was thinking that by using the google tool, you would avoid any slap action for duplicate content from google, since they know you are testing splits. Thks.

From conversion optimization point of view, we usually launch with one page for at least a week or two [depends on the amount of traffic you receive or pay for]. Our goal is to establish a control data set. Then we move into conducting either A/B or mutlivariant testing. GWO works wonderfully for both.

As far as long vs. short copy impact on quality score, I have many short copy pages with high quality score.

[crocwatcher]

cool, thks!

[chris1]

Originally Posted by crocwatcher

You are correct Chris, I am referring to landing pages, not ads. So you would not use a/b split tests inside of google's web site optimizer then? Have you had any bad experience with using google's tool for this? I was thinking that by using the google tool, you would avoid any slap action for duplicate content from google, since they know you are testing splits. Thks.

I suppose you could use the optimizer, though its actually easier for us just to test ourselves. Also, I wouldn;t worry about being penalized for having the same term in two seperate campaings/ad groups... the only penalty you get for that is that only one ad will show at a time (the one with the higher bid, quality score, CTR%, etc.).

[crocwatcher]

hey Chris, I wasn't worried about a given term but > 50% duplicate content type of slap. If you are split testing 2 landing pages and the content is very similar, I was concerned about duplicate content banning, which I have encountered in the past. My thinking was that if you use the optimizer, google would give you slack on this issue since they know you are split testing... thanks for you input and help!

[khalidh]

Originally Posted by chris1

I suppose you could use the optimizer, though its actually easier for us just to test ourselves. Also, I wouldn;t worry about being penalized for having the same term in two seperate campaings/ad groups... the only penalty you get for that is that only one ad will show at a time (the one with the higher bid, quality score, CTR%, etc.).

I guess you can always test the pages yourself if the only thing you are trying is true A/B test. However, sometimes there is a lot more to the math than calculating and comparing CR for two different pages. This is of course is more true in the case of multi-variant testing. When you utilize GWO, there is no duplicate content issues since you will be testing one page with two different versions controlled via javascript.

[chris1]

Duplicate content should not an issue in paid campaigns.

[crocwatcher]

thanks folks!