You can create a video experiment to determine which of your video ads is more effective on YouTube. With a video experiment, you can test different video ads with the same audience, and then use the results of the experiment to determine which ad resonates more with your audience. This article explains how to create a video experiment.
There are two video experiment paths, basic and advanced. The basic option is called A/B test video assets, where creative is your single variable in your A/B two-armed test (control vs. treatment). The advanced option is called Custom, where you can test any variable and any number of variables up to 10 arms.
At this time, A/B test video assets are only available for Video Reach Campaigns (Efficient Reach) and Video View Campaigns. For all other video campaign types, use Custom. For Performance Max (PMAX) campaigns, go here. For Demand Gen campaigns, go here.
Basic option: A/B test video assets
Overview on how it works
- Select a success metric to test
- Absolute Brand Lift
- Clickthrough rate (CTR)
- Conversion rate (Conv. rate)
- Cost per conversion (Cost/conv.)
- Cost-per-click (CPC)
- Cost-per-thousand impr. (CPM)
- Cost-per-view (CPV)
- Video view rate
- Select one Video View Campaign or Video Reach Campaign (Efficient Reach) campaign to serve as your control arm.
- The system will allow you to duplicate the selected campaign (the duplicate is your treatment arm) and you can add or remove video assets to understand asset level impact on your campaign (for example, CPV, Views, VTR, Brand Lift, and more).
- Make a data-driven decision on which creative assets drive the best campaign outcomes for your goal.
Example
A car manufacturer wants to check which of their video ads for a new sport utility vehicle (SUV) leads to more conversions. The manufacturer is targeting users in the market for a new luxury or performance vehicle. They set up an experiment to compare the performance of 2 video ads (one in each campaign) and choose “Conversions” as their success metric. After running the experiment, they find that one ad outperforms the other ad in conversion volume by as much as 8 times.
Before you begin
- Start with a hypothesis to determine your testing strategy. Your hypothesis should be tied to your overall goal for the campaign. For example: “Which of these video ads drives the higher view-through-rate for my customer acquisition campaign: a 2-minute tutorial or a 15-second direct-offer video?”
- Create an awareness and consideration Video campaign (Video Reach or Video Views) before setting up an experiment (this will be your control arm). Learn more about how to Create a Video campaign.
- As you create the experiment, the system will enable you to duplicate the campaign (this is the treatment arm) using the same settings (such as audiences, bids, ad formats, and more) and try different creatives to understand how each video ad creative performs independent of other campaign characteristics.
- Easily add or remove video ad creatives in the treatment arm to understand the impact (for example, CPV, Views, VTR, Brand Lift, and so on) that the assets had on your campaign’s performance.
- Keep in mind that Brand Lift is not available for all Google Ads accounts. To measure Brand Lift in a video experiment, contact your Google account representative. If you don't have a Google account representative, you won't be able to measure Brand Lift in the experiment.
Instructions for A/B test video assets
Create your video experiment
The following is only available for single variable creative tests for Video Reach Campaigns and Video View Campaigns:
- In your Google Ads account, click the Campaigns icon
.
- Click the Campaigns drop down in the section menu.
- Click Experiments.
- Click Video experiments.
- If you are A/B testing creative as a single variable, select A/B test video assets.
- If you are testing audiences, bidding strategies, formats or creative with more than two experiment arms, select Custom (advanced) and proceed to the Instructions for Custom video experiment.
- Click the Select a metric drop-down and select one success metric among:
- Absolute brand lift
- Clickthrough rate (CTR)
- Conversion rate (Conv. rate)
- Cost per conversion (Cost/conv)
- Cost-per-click (CPC)
- Cost-per-thousand impr. (CPM)
- Cost-per-view (CPV)
- Video view rate
Notes:
- If you encounter warning boxes during the experiment setup process, there are potential issues to address. Please resolve these issues before saving the experiment.
- Absolute Brand Lift is not available for all advertisers.
- Click Select campaign to choose your control.
- A/B test video assets is currently only available for Video Reach Campaigns (Efficient Reach) and Video View Campaigns so only these campaigns will appear in your list of options. If you would like to test another video campaign type, select Custom (Advanced). For Performance Max (PMAX) campaigns, go here. For Demand Gen campaigns, go here.
- If you select Cost per Conversion or Conversion rate, you must have conversion tracking set up in your ad group.
- When choosing Brand Lift for your experiment, you'll need to create a Brand Lift study within the selected campaign for your control group before starting. You'll be prompted to do this during experiment creation, and the warning will only disappear once you've set up a study and taken additional steps in the video experiment setup (such as saving, modifying assets, or duplicating the campaign). Learn more about how to Set up Brand Lift measurement.
- Click Duplicate your control campaign (this is now your treatment campaign).
- At this stage, the treatment campaign will be identical to your control campaign. In the following step, you'll replace the ad creatives (your variable) in the treatment campaign.
- Click Add videos, to add or remove videos within your treatment arm.
- Review your experiment budget and experiment dates.
- Name your experiment.
- Click Save.
Note: Apart from ad creatives (your variable), any adjustments you make to the control campaign's settings (such as budget, flight dates, and so on) will automatically be applied to the treatment campaign during the experiment. When your experiment finishes, you'll need to manually end it.
End your video experiment
When an experiment is running, traffic is split evenly between the campaigns. Before you make changes to your campaigns, you should end the experiment. After you end the experiment, the campaigns will return to their previous traffic distribution levels.
- In your Google Ads account, click the Campaigns icon
- Click the Campaigns drop down in the section menu.
- Click Experiments.
- Click Video experiments.
- Find the experiment that you want to end, and click End experiment to stop your treatment campaign.
- For A/B testing:
- If your treatment campaign was more effective, choose the first option to replace your existing control campaign with the assets from the treatment campaign.
-
If your control campaign was more effective or if you don't want to apply the treatment campaign's assets to your control campaign, choose the second option to end the experiment without applying assets.
Advanced option: Custom
Overview on how it works
- Select a success metric to test:
- Absolute Brand Lift
- Clickthrough rate (CTR)
- Conversion rate (Conv. rate)
- Cost per conversion (Cost/conv.)
- Cost-per-click (CPC)
- Cost-per-thousand impr. (CPM)
- Cost-per-view (CPV)
- Video view rate
- Select one video campaign to serve as your control arm.
- Select your campaign(s) for your experimental arms. By default, there are 2 experimental arms, add more if needed (up to 10).
- Make a data-driven decision on which video ads drive the best campaign outcomes for your goal.
Before you begin
- Start with a hypothesis to determine your testing strategy. Your hypothesis should be tied to your overall goal for the campaign. For example: “Which of these audiences drives the higher view-through-rate for my customer acquisition campaign: Males 18+, Female 18+, Males 25-54 or Females 25-54? For example: “Which of these creative campaigns drive the highest brand awareness lift for my outdoor enthusiast target audience: Fishing, Camping, Scuba Diving or Hiking?
- Create your video campaigns before setting up an experiment.
- Select the campaigns you want to test as your experiment arms.
- We recommend one variable (such as audiences) to keep the experiment clean.
- Keep in mind that Brand Lift is not available for all Google Ads accounts. To measure Brand Lift in a video experiment, contact your Google account representative. If you don't have a Google account representative, you won't be able to measure Brand Lift in the experiment.
Instructions for Custom video experiment
- In your Google Ads account, click the Campaigns icon
- Click the Campaigns drop down in the section menu.
- Click Experiments.
- Click Video experiments.
- Select Custom (advanced).
- Click the Select a metric drop-down and select one of the following success metrics:
- Absolute brand lift
- Clickthrough rate (CTR)
- Conversion rate (Conv. rate)
- Cost per conversion (Cost/conv)
- Cost-per-click (CPC)
- Cost-per-thousand impr. (CPM)
- Cost-per-view (CPV)
- Video view rate
Notes:
- If you encounter warning boxes during the experiment setup process, there are potential issues to address. Please resolve these issues before saving the experiment.
- Absolute Brand Lift is not available for all advertisers.
- Select one campaign to serve as your control arm (must set up campaigns prior to experiment set up).
- Select campaigns for your treatment arms (up to 10 arms total).
- Follow the prompts to adjust traffic split.
- Review your experiment budget and experiment dates.
- Name your experiment.
- Click Save.
End your video experiment
When an experiment is running, traffic is split evenly between the campaigns. Before you make changes to your campaigns, you should end the experiment. After you end the experiment, the campaigns will return to their previous traffic distribution levels.
- In your Google Ads account, click the Campaigns icon
- Click the Campaigns drop down in the section menu.
- Click Experiments.
- Click Video experiments.
- Find the experiment that you want to end, and click End experiment to stop your treatment campaign.
- For A/B testing:
- If your treatment campaign was more effective, choose the first option to replace your existing control campaign with the assets from the treatment campaign.
-
If your control campaign was more effective or if you don't want to apply the treatment campaign's assets to your control campaign, choose the second option to end the experiment without applying assets.
Directional vs. Conclusive Results
You can view the supposed winner of your experimental arms while the video experiment is still running and gathering data, allowing you to have directional results earlier. We recommend waiting until the experiment finishes for conclusive results but if you don’t have time and are comfortable with directional results, you can have them as early as when the experiment reaches a threshold of 70% confidence. You also have the option to wait until the experiment reaches 80% confidence (still considered directional results) or wait until it reaches completion at 95% confidence (results at this stage are considered conclusive results).
A threshold of 70% confidence or what is known as a “confidence interval” means that if you were to repeat this experiment, you would get this same result 70% of the time. If you choose to wait for conclusive results, those will be at a 95% confidence interval.
Terms for video experiments
You may not be familiar with some of the terms for video experiments. To familiarize yourself, expand the section below to find a list of terms and their definitions.
Expand all
- Hypothesis: A hypothesis is a statement or question that your experiment will answer.
- Experiment arm: An experiment arm refers to a campaign or a group of campaigns that have specific settings. Video experiments have at least 2 experiment arms so that you can compare performance between them.
- Traffic split: The traffic split is how you want to split your experiment. Most experiments with 2 experiment arms have an even traffic split of 50% per arm.
- Statistical significance: A determination that a relationship between 2 or more variables is caused by something other than chance. In other words, when an experiment has reached statistical significance, you can trust the results are based on more than just random chance or luck.
Next steps
Monitor the performance of a video experiment
