How To Create An Experiment (Google)
One of the most useful features for paid advertising is the ability to perform A/B experiments. Platforms like Facebook, LinkedIn, Bing, and Google Ads allow you to do so in order to test your creatives for anything you want! You can perform an experiment to test which landing page would work best for your ads, various bid strategies, headlines, descriptions, etc. The freedom is all yours! The purpose of testing is to find what works better between two variables, and continue with the winning strategy to improve campaign performance. I will walk through how to create an experiment on Google.
Step 1: You will be asked to name the experiment. Tip: Name it in a way that when you look back you know what it’s for.
Step 2: You will be asked to choose a base campaign that you’d like to use for your experiment. Afterwards it will ask you to again name your trial campaign in which you are running the test to distinguish it from the original campaign.
Step 3: You will be asked to choose a goal that you are testing for:
- Clicks → Maximize Clicks
- Impressions —> Target Impression Share
- Cost
- Cost Per Conversion → Target CPA
- Conversion Value Per Costs → ROAS
- Conversions → Maximize Conversion
- Conversion Value → Maximize Conversion Value
Step 4: You will be asked in relation to the metric chosen, if you wish to view an increase, decrease, or no significant change.
Step 5: Here you will decide how much of the budget/traffic will be driven to the experiment. Best practice is to do a 50/50 split in order to provide the best comparison between the original and experiment campaign.
Following, it will ask you how you’d like to select a split option: search-based or cookie-based. Google recommends choosing cookie-based to ensure your split is as fair as possible, so that anytime a user is doing a search, Google will only show one variation of the ad to the user. A search-based split means that anytime a user makes a search more than once, they could be served ads from both the Base and Experiment campaign. This tends to be less accurate, but will gather more data in a shorter amount of time.
Step 6: Set the date you want the experiment to start. Google gives you an option to choose an end date. You have the freedom to choose to end the experiment whenever you want or let it go on without a specific time. If you do not have a specific date that you would like to end your experiment, our recommendation would be to choose “None” so that you can run the experiment as long as you need to and end it in real time whenever you’re happy with the results, or you can set the end date later.
Step 7: You have the choice of enabling sync for your campaigns you are running the test for.
Use case example: New keywords were added to the base campaign. Google will automatically update those changes within the experiment you are running so you don’t manually have to do so. Sync is on by default when you begin a new experiment.
That’s it! You have just created an experiment! Sit back and relax while Google does the experimenting for you.
Comments are closed here.