• Diseño web Galicia | @matpsic

    +433%! and -80.4% CPA that is awesome. I will try it.

    Thanks a lot.

  • Mark Kennedy

    Another good post, Bonnie. I think a lot of people will benefit from this step-by-step. Out of all of these steps, I think #6 is something to really watch. One change to the control group not made to the experiment group could really throw off the data. Everything has to be consistent, except the issue you are testing. However, #9 is a great feature, that makes PPC life easier if you have killer results.

  • Lani Kee

    Aloha Wil, very cool tips you have displayed on your post. Enjoyed viewing it and learning from it. I may consider viewing what else you have on your blog. I as well like your video clips. Thanks for sharing!

  • Bonnie Schwartz

    @Mark Like always thanks for your awesome comment. They are always greatly appreciated! #6 is definitely key. The reason this worked really well for this example was again because it was a completely new campaign. I was very cognizant of mirroring any changes to the control group and experiment group. I had good positions across the board so I was bit more hands off on the campaign than I normally am as I gathered results. I could not help myself in making some changes like adding negatives based on search query reports that gave me angst :) but again I did that at the campaign level. I loved that I was able to see the results right in the interface. I really want ACE to roll out to campaign settings because I would love to test day parting, site links among other things with ACE. Usually when I wish for something the brilliant team at Google eventually grants my wish so time will tell.

    @Lani Glad you liked the post!!

  • Bonnie Schwartz

    @matpsic Good lucking Testing!

  • Rob Kingston

    AFAIK, I can’t think of any better way to test multiple offers on landing pages without having to compromise the ad copy in AdWords.

    As for #6 – it’s important until you want to test how effective your last minute changes are on the campaigns – could be good to show a measureable improvement – even if CPCs continue to rise.

  • Bonnie Schwartz

    @Rob Thanks for your comment! Good point about #6. I just added your blog to my Google Reader. Some great GA tidbits!

  • Jeremy Yandell

    Hi Bonnie,

    I was wondering why you set up the experiment with a control and an experiment ad group rather than just stick to testing at the ad level? Couldn’t you just keep a single ad group and have ad A with the old Landing page as the control and ad B with the new Landing Page as the experiment? That way you wouldn’t have to make bid changes, keyword changes etc in both ad groups.

    (I did notice this was an older post and maybe this was just the only way to do it since ACE was new.)

  • Bonnie Schwartz

    @Jeremy Yep! When ACE 1st came out this was not possible. I wrote a follow up post on how I would set this up now:

  • art

    Hi Bonnie,
    It doesn’t look like ACE provides a way to test landing pages only without testing the adcopy.
    If you were only interested in testing the landing pages, website optimizer might be the best fit.
    what are your thoughts?

  • Bonnie

    @Art The Website Optimizer definitely is still a strong alternative. As no special programming is needed. If you want to to test more than one Landing Page or for multivariate testing the Website Optimizer is still your best bet

  • Bonnie

    @Art I wanted to make sure to clarify my comment above as I hit submit before finishing editing. The Website Optimizer is still a great alternative. However, it requires special programming, which sometimes can slow down the process if tech resources are limited. With ACE no special programming is needed, so it is a quick way to test two separate landing pages for your ppc traffic. If you want to to test more than one Landing Page or for multivariate testing the Website Optimizer is still your best bet.

  • Robert Kingston

    @Art & Bonnie – if you’re afraid of the technical aspects of Google Website Optimizer, you can always try the easy to use Visual Website Optimizer. Checkout a bunch of reviews of split testing software – there’s some really great products out there now, but GWO and VWO are my favourites.

  • Zac Stafford

    So can you clarify something for me? Are we supposed to have one campaign with two different adgroups (one named ACE and the other the existing adgroup)?

  • Bonnie Schwartz

    Hey Zac,

    Actually this post is a bit out dated as Google updated ACE so that you can use experiment ads, which was not live when I wrote this. I would use this approach now: