A while ago I wrote a blog introducing the beta of Google Campaign Experiments. Well a couple of months have passed and I have been experimenting away on some campaigns and thought I would share some tips/ideas with you all.
Basically for those who haven’t heard about this before then in a nutshell Campaign Experiments let you run split tests on different aspects of you PPC campaign. This way you can run tests without other factors such as seasonality, competitor activity, or any other online or offline marketing that may occur. Here is a brief guide on how to set up an experiment:
Go to Campaign Settings on the campaign you want to run your experiment on and under Advanced settings there should be an ‘Experiment’ option.
Click the ‘Specify Experiment Settings’ button then enter the values for experiment parameters.
Here you can name your experiment and select the frequency that the experimental change is applied (by choosing the % split between ‘control’ and ‘experiment’ which can help minimise the risk if you are making significant changes.). You can also define how long the experiment should run from and until. Now just click save and it’s time for the fun stuff.
After you’ve set your experiment parameters, you need to decide what kinds of experimental changes you’d like to test. You can make experimental changes to keywords, ad groups, or ads. Go to one of those tabs, and make the changes you’d like to test with your experiment.
To add new keywords to the experiment split go to the Keywords tab and click the Add Keywords button. Add the keywords you’d like to add as you normally would then select the checkbox that reads, “Save these keywords as experiment-only.” For Ad Groups it’s the same, just go to the ‘Ad Group’ tab and select ‘Add Ad Group’.
You can also use experimental bid changes. Simply go to the Keywords tab and segment by experiment and on the experiment row of a keyword, click on the bid. Here you can add or subtract a percentage from the control bid which will create a new experimental bid. The same goes if you’re making ad group bid changes.
So once you get to grips with the tool and how to set up your experiment the next step is to choose what kind of experiment to run. Here are a few ideas to get you started:
You can use ACE to test the impact of incremental bid changes, and you can make wholesale bid increases or decreases across your experimental campaign. Ever wondered what would happen if you pushed those keywords above the fold? Now you can, but without risking blowing all of your budget!
Why not launch those new terms you’ve been mulling over adding as an experiment? You’ll be able to see how your ad groups perform with and without these new keywords so you can make a truly informed decision. You could even try removing some keywords, will this drive more traffic through your other terms? There’s only one way to find out…
Why not try more granular Ad Groups and test to see how this impacts your performance? Or maybe you want to see what would happen if you try a slightly different theme? Just follow the steps in the guide and set up your new ad groups as experiment only whilst keeping your originals as control and away you go!
You could try out phrase or broad match on some of your keywords, just create an experimental ad group with these new match types in, and run your original ad group as control only. Now you can see a side by side comparison of how these new match types affect the performance of your ad group, without the risk of undoing any or your hard work.
There have been a few blogs about this so I can’t take this as one of my own ideas but I think it’s a pretty nifty use of ACE. I have taken the below from a blog that Brad Geddes wrote (here). This is his process for testing keyword suggestions from Google Instant and it’s rather interesting to use this relatively new tool to test some of the impact of Instant:
- Choose a few of your root keywords (generally the 2 word variety)
- Type these words into Google instant or web seer
- Make a list of these words
- Break them into appropriate ad groups
- Enable ACE (AdWords Campaign Experiments)
- Create new ad groups with these words
- Make these brand new ad groups part of your experimental campaign
- Let the experiment collect data
- Measure the results
Might be worth a try if you have some budget to test with…
OK so we’ve gone through the set up and also some ideas for experiments, so I thought I would briefly cover some minor gripes users may have with ACE.
Firstly and in my opinion the main gripe with this tool is the reporting. It’s not possible to look at the overall performance of control against experiment unless you download the data and perform some jiggery pokery yourself. Also, when exporting reports the download is not split by experiment or control, just the usual overall report, which makes manual review outside of the adwords interface a little troublesome.
Secondly you can only run one experiment at a time on each campaign, it’s not a huge deal and shouldn’t really hinder you as this enables you to see the effects of a certain experiment fully, rather than muddy the results with multiple variables.
Finally, and this more of a wish than a gripe, there currently isn’t support for ACE in AdWords Editor so all changes must be made in the web interface.
So there you have it, Google Adwords Campaign Experiments in a (rather large) nutshell. I think this tool really has legs and with intelligent use can really help to fine tune PPC campaigns. Have you tried ACE? If so I would love to hear your thoughts, whether they are shining endorsements or damning dismissals!