In order to get the best results from your PPC campaign, it is important to always test new things. Test keywords, ads, settings, budgets, keyword bids, etc. – if it is something you can change, you should test a variation. The trouble is, it is often difficult to determine if the change you made is really a success or if other factors are contributing. For instance, maybe you increased bids and saw an increase in conversions – we can assume the bids are the reason for the boost in conversions, but maybe your competitors pulled back their budgets allowing you to capture more of the market so really your bid adjustments didn’t have anything to do with it. There hasn’t been an easy way to isolate changes and test performance to see if one change really makes a difference, until now.
Google AdWords has been working on new features to help manage tests with an initiative called Adwords Campaign Experiments or ACE. It is still in Beta, but is now available for some users to test out for themselves. If this is something you want to join, simply submit your AdWords Customer ID for consideration. You won’t get an email saying you have been granted access, so check your Settings tab after a few days to see if the option has been added. If access has been added to your account, it will be under Advanced Settings and labeled as “Experiment”.
So, you were able to get access to the new tool – now what? Get ready to start testing! The Experiments feature works on existing campaigns, so you can make adjustments and compare it to your control group. To get started, click on the Experiment link in the Setting tab. You will need to enter a name for the test, start date and end date. The default testing time period is 30-days, so make sure you have enough traffic coming into the campaign to provide results in that time period. If your traffic levels are lower, consider extending the test for a longer time period to make sure you get enough data to make an accurate conclusion. You also can select the percentage of exposure directed to your experiment vs. the control campaign. The default is a 50/50 split, with 50% of the time your control being used and the other 50%, your test. This is the recommended setting but if you are testing something significant that may greatly change your overall spend or performance, you may want to allocate slightly less towards the experiment. For instance if your experiment is based on increasing your keyword bid by 60%, you may want to direct a smaller portion of your budget to your experiment in order to help control overall spend.
At this time you can experiment with keyword bids, keywords, ad text, ad groups, keyword match types, ad group bids and keyword insertion. Within the content network, you can also test bids on managed placements, new placements, additional keywords, ad text or display ads, new ad groups, and keyword bids. Basically, you can test anything besides campaign settings. Your current campaigns settings including geo-targeting and ad rotation will be applied to your test so everything is consistent.
Once you set up your experiment on the settings tab, you can now adjust the elements you want to test. Make sure you think about your strategy and approach prior to making any changes so you will gain valuable information from the experiment. Be sure you are only testing one element per experiment so the data is statistically significant and can be attributed to a specific change.
Let’s say for your experiment, you want to bid up on keywords to see if a higher placement will get you additional conversions. To do this, go to the keywords tab within the campaign that has the experiment set and you should see all of your keywords with two rows below. One row allows you to set a bid for the Control and one for Experiment. If you do not see this, go to the Segment button (just below the Settings tab) and select “Experiment” – this will bring in the additional rows needed to manage bids separately. To adjust bids, click on the bid in the experiment line item and then adjust either up or down by the percentage you want it to change by. For instance you may want to see if you get additional conversions by setting your bids 20% higher than the control.
Maybe instead of keyword bids, you want to test a new set of keywords against ones currently in the account. To add these in, go to the Keywords tab at the top. Add in the keywords as you normally would, BUT make sure you select the “Add as experiment only keywords” check box at the bottom before hitting save. This will put the new keywords into the experiment group and will not add them to the control group.
There are a lot of different options you can choose from when setting up your experiment – try clicking on the beaker to the left of the keyword and you can determine if a keyword should run in only the Control group, only the Experiment group or both. Again, be sure not to test more than one thing at once, you can always run additional experiments later to see if other changes positively impact your account.
Your test it set up and running, now what? Watch and observe. Try not to make any adjustments to that campaign during the time period in order to make sure the data you are gathering is accurate. You will see in the data lines, small arrows that show the difference between the control and experiment. If data is statistically significant you will see multiple arrows to help flag the information.
Your experiment will automatically end on the date you selected, so there is no need for you to do anything more except watch the account. Depending on the change, you may find that the data you are seeing is statistically significant in a shorter amount of time so make sure you are checking back.
At the end of your experiment you have to option to apply the experiment changes to the campaign. Let’s say you now know that increasing your bids by 20% brings in significantly more conversions and you want to apply the changes made in the test to the campaign long-term. To apply the changes exactly as they are, go to the Settings tab in the advanced settings where you initially set up the experiment. Select the “Apply: launch changes fully” button and the changes you made will be applied. If you determined you do not want to make these changes but rather go back to your original settings for all search queries, select the “Delete: remove changes” button. One note, once you delete the changes you will no longer have access to the experiment data so be sure you have what you need prior to selecting this.
At the end of the day, the tool is a welcome addition to the AdWords settings, even in Beta format. If you are able to gain access to it I strongly recommend testing – you might be surprised at the results.
