Google Ads draft and experiments setup allows you to test changes against your existing campaign in a controlled split-traffic test to measure performance impact. By creating a draft, you can prepare adjustments to your campaign settings, bidding strategies, or ad creative without affecting the original campaign’s live performance. This method provides a clear way to see if your proposed changes actually lead to better results before you commit to them permanently. Using this structure helps advertisers avoid guessing whether a change will work, relying instead on direct evidence from their specific audience.
Why should you use campaign experiments?
Testing is a core part of digital advertising because what works for one account may not work for another. When you use experiments, you split your campaign traffic, meaning that a portion of your users sees the original campaign while another portion sees the experiment. This split ensures that your data remains clean and comparable. You can track performance metrics such as conversion rates, cost per acquisition, or click-through rates side-by-side. This approach is useful when you want to change your bidding strategy, such as switching from manual CPC to Target ROAS, or when testing new landing pages.
How do you create a draft and experiment?
To start, find the ‘Experiments’ tab within your Google Ads account interface. You will first create a draft, which acts as a copy of your chosen campaign where you can implement your desired modifications. Once the draft is ready, you initiate the experiment, which divides your traffic between the original and the new version. You must define a start date, end date, and the percentage of traffic you want to allocate to the experiment. In a typical account setup, a 50/50 split is common for statistical significance, though you can adjust this based on the volume of traffic your campaign receives.
- Go to the Campaigns page and select the campaign you want to test.
- Navigate to the ‘Drafts & experiments’ section in the left menu.
- Select ‘New draft’ and give it a descriptive name.
- Apply the changes you wish to test, such as updating ad copy or bidding.
- Click on ‘Experiment’ to define the traffic split and duration.
How to evaluate your test results
After the experiment runs for a period, you need to check the results in the Experiments dashboard. Google provides a comparison view where you can see the performance of the base campaign versus the experiment. Look at the confidence intervals to determine if the differences in performance are statistically meaningful. If the experiment shows a positive impact, you have the option to apply those changes to your original campaign permanently. If the results are neutral or negative, you can simply end the experiment without having disrupted your primary campaign’s performance.
Comparison of Testing Methods
| Feature | Campaign Draft | Campaign Experiment |
|---|---|---|
| Primary Purpose | Change preparation | Performance validation |
| Traffic Exposure | Zero impact | Split traffic |
| Data Collection | None | Real-time comparison |
How Alien Road applies experiments
At Alien Road, we follow a standard process for managing client experiments. First, we identify a specific variable to isolate, such as testing a new automated bidding strategy against a manual one. We ensure the campaign has enough historical conversion volume to provide meaningful results. Then, we set up the draft and monitor it daily to ensure no technical errors occur. After the testing period ends, we present a report to the client detailing the impact on key performance indicators. This systematic approach allows us to make informed decisions that benefit the long-term health of our clients’ ad accounts.
What are common pitfalls in setup?
One common issue is running an experiment for too short a time, which prevents the algorithm from gathering enough data. You should ensure the experiment runs long enough to account for weekly cycles in consumer behavior. Another error is changing too many variables at once; if you change your audience, your bidding, and your ad copy in one experiment, you will not know which change caused the results. Keep your experiments focused on a single change to get the most accurate data possible.
- Avoid testing too many variables simultaneously.
- Ensure your experiment runs for at least two to four weeks.
- Check the experiment status regularly for any delivery warnings.
- Monitor budget allocation to ensure the experiment doesn’t exhaust funds.
Where to start this week
This week, identify a campaign in your account that has consistent daily spend and stable conversion performance. Create a draft to test one specific change, such as a new landing page URL or a change in ad headlines. Set the traffic split to 50% and let the experiment run for at least two weeks. By the end of the month, you will have a clear answer on whether your proposed change is worth implementing. This proactive method helps you maintain control over your account performance while continuing to seek growth.
Frequently Asked Questions
Can I run multiple experiments at once?
Yes, you can run multiple experiments simultaneously, but it is not recommended if they are testing the same variables. Running overlapping tests can lead to data contamination, making it difficult to attribute performance changes to a specific factor. Keep your experiments distinct to ensure each test provides clear and actionable insights for your account.
Will experiments affect my Quality Score?
Experiments are designed to isolate the performance of specific changes. While the experiment itself will have its own data, it should not negatively impact the Quality Score of your base campaign. The base campaign continues to function independently, and the traffic split simply diverts a portion of your users to the experimental version for the duration of the test.
How long should an experiment run?
A typical experiment should run for at least two to four weeks. This timeframe is necessary to capture enough conversion data to reach statistical significance. If your campaign has very low traffic, you might need to run the experiment for longer to gather sufficient data points, but avoid keeping experiments open indefinitely as this creates unnecessary account clutter.
What happens if I stop an experiment early?
If you stop an experiment early, you simply lose the chance to gather more data on your test variable. The experiment will end, and your campaign will revert to running only the base version. Stopping early is fine if you notice the experiment is causing a significant decline in performance, but try to avoid it if possible to get clear results.
Keep reading: Google Advertising · Advertising