ACE Tests: It’s Just Day-Parting. Really Fancy Day-Parting.

By Nicole Gardner, Paid Search Specialist, Advertising Solutions

You may have heard of Google’s ACE (AdWords Campaign Experiments) tool, allowing for advertisers to test creative, keywords, bids, match types and anything else that can be changed at the ad group or keyword level.

The tool is great, but the way Google presents it can make it seem very complicated. Good news: it’s actually pretty simple. I’ve run some tests and found it much easier than expected. For example, in one account I found that lowering keyword bids by 20% also lowered the position, CTR and CPC by about 20% (not a shocker). However, in another account this same action lowered average CPC by 20% but lowered position and CTR by less than 5%!

Don’t you want to know if you can lower your CPC without sacrificing CTR? Yes? Then allow me to enlighten you.

The Basics to ACE
ACE is really just a day-parting tool. Honestly, that’s it. It sounds complicated because you have to designate some keywords “control” and other keywords “experiment” – and maybe this makes you want to throw up your hands because you never took AP Chemistry in high school. No worries – think of it this way: When you set up an ACE test in the Campaign Settings tab, you are designating ad serving.

Once you set up the test with end dates and ad serving, you have to designate the keywords/ad groups according to how you want them day-parted. Let’s use 70%/30% ad serving as an example for below.
  • “control and experiment” means these keywords/ad groups will show 100% of the time because they aren’t part of the test.
  • “control only” means these keywords/ad groups will run 70% of the time.
  • “experiment only” means these keywords/ad groups will run 30% of the time – only when the “control only” keywords/ad groups are off.
Keyword Testing
Say you’re testing a keyword on broad versus on phrase. If you already have the keyword on broad match, then add the keyword to the same ad group as phrase match. Set the broad match as “control only” and the phrase match as “experiment only” and all other keywords as “control and experiment.” Then go back to the campaign settings and “start running experiment.” It will run until the end date you set.

Ad Group Testing
Before you “start running experiment,” go to Editor and copy/paste the ad groups you want to test, then append “experiment” or “test” to the new ad group names. Change what you want to test – for example, change the ads to have a slightly different display url. Upload these new ad groups as paused. Then, after you set up the end dates and ad serving in the experiment, designate the new ad groups “experiment only” and the original ad groups “control only.” Everything else is “control and experiment.” Then go back to the campaign settings and “start running experiment.”

Last Points
Another confusing aspect of ACE are the buttons that appear in the Campaign Settings that say “Apply: Launch changes fully” and “Delete: Remove changes.” “Apply” means that everything you designated as “experiment only” will now run 100% of the time. “Delete” means “control only” will now run 100% of the time. Don’t touch these until you are done with your experiment. Before hitting “Delete,” download the history of the test into Excel so you can have the results of your test, otherwise you’ll lose it. Also, try to keep your control ad groups and experiment ad groups as similar as possible other than what you are testing. For example, if you are testing ad copy changes, you want to make sure the keyword bids in both ad groups are exactly the same.

Hope this helps. Start testing!


Copyright © 2008 Resolution Media, Inc. All rights reserved.