Now you can test your search and display network with a google ads new split testing feature called draft & experiment. Now you don’t need to buy or use another external tools to run a A/B testing.
Draft: You can use draft feature to make multiple changes to your campaigns without disturbing live campaign. So you can apply your draft changes to your live campaign or you can a run a experiment with it.
Experiment: You can run a slit campaign test to understand which performance better without making any changes on the live campaign, Once the campaign is over you can apply them on live campaign.
Few things you should know about Google Ads Draft & Experiment.
- You can not pause your Google Ads campaign experiments: Google ads experiments don’t have a pause option. You can stop this only by adjusting the experiment “End date” or you can discontinue with the experiment.
- It works only on search & display networks campaigns. As of now it’s not available for Video, App or Shopping campaigns.
- Completed or deactivated experiments can not be activated again, you need start a new experiment.
- Be patience, Google Ad experiment need a learning period: You need have a patience while running a experiments, it takes 10 – 15 days time for learning period because, it does not take any historical data from the original campaign, so basically starts from scratch, it create its own learning graph. It will take few weeks to learn, So you need have a little bit of patience and don’t come to the conclusion or pause the experiment within few days.
- One Experiment per Campaign: You can run only one experiment at a time but you can schedule upto 5 experiment per campaign.
- If your campaign is under shared budget than you can not create a new experiment.
- Use the experiment end date bit further because, it wont allow to extend the experiment date once it go live. But you can always conclude the test early.
- Launch the experiment at least with 50% traffic split to get enough traffic for each to test which works best.
- During the experiment period, do not make any changes that reset the quality score on your original campaign, it will affect your experiment campaign.
- Experiment works best for the campaigns with high traffic volume.
- Once the experiment is over, if you are happy with the results than you can apply your experiments to original campaign or you can convert into new campaign.
- There are two type of experiment treatments you can use, search-based split and cookie-based split.
Search-based split: user are randomly placed in either original or experiment campaign every time a same user search query occur.
Cookie-based: Here user can see only one version of the campaign ( Search-based split or Cookie-based) regardless of how many time they search query comes from same person.
- For search campaigns you can use cookie-based split or a search-based split.
- In Display campaigns, always use cookie-based split.
You can use Google ads Experiments to test different campaign settings like
- Automated bid strategies
- Device settings
- Target CPA
- Ad extensions
- Negative keywords
- Structural campaign changes
- Ad rotation
- Different landing page design
- Demographic adjustments
Display & Remarketing targeting:
- Remarketing audiences
- Display interests
- Display placements
- Display topic
According to Google Features and reports that aren’t available for drafts (Source: https://support.google.com/google-ads/answer/6318732?hl=en)
- Ad schedule report
- Category & Search terms
- Auction Insights
- Display Placements report
- Scheduled email reports
- Bid landscapes
- Ad customizers that use “Target campaign” or “Target ad group”
- Keyword diagnosis
- Some automated bid strategies:
- Target search page location
- Target outranking share
This is one of the best feature introduced by Google ads, you can run a A/B test experiment without disturbing your present campaigns and it’s very easy to analysis and implement best performing campaign. If you have any questions or require any Google Ads consultation than please feel free to get in touch with me.