Profile
Dashboard
Logout
Table of contents
Key takeaways
- CompleteGreet splits visitors evenly between Video A and Video B inside one single video bubble.
- Change one part of Video B, such as the opening line or call to action, while keeping the bubble settings the same.
- Wait for enough traffic and 95% statistical confidence before choosing a winner.
How to A/B Test Video Bubbles on Your Website
CompleteGreet can split visitors evenly between Video A and Video B inside one single video bubble. Each visitor keeps the same version during the test, and the analytics dashboard compares bubble open rate before it suggests a winner.
Most teams mess this up by changing too many variables at once.
Use the same page, placement, trigger, buttons, speaker, framing, and video length. Change only one part of Video B, such as the opening line or call to action. CompleteGreet handles the 50/50 split automatically.
Watch for seasonal noise.
Campaigns, launches, holidays, or site changes can distort results. Let the test cover at least one complete weekly traffic cycle, then keep it running until the dashboard has enough visitors. If traffic is low, the calendar alone does not make the result reliable.
CompleteGreet uses bubble open rate as its decision metric. The dashboard also shows full views, button clicks, form submissions, and average watch time. Decide before the test which business result you care about, but do not choose a winner from whichever number looks best afterward.
Keep Video A and Video B similar in length. The editor warns when the gap is more than 5 seconds and blocks the test when the gap is more than 10 seconds. Check both versions on a real phone before publishing.
Why A/B testing video bubbles matters
A/B testing replaces a guess with a direct comparison. You can see whether a different opening line or message changes how often visitors open the bubble.
CompleteGreet’s published cases show results from full implementations, not from one combined A/B test. Merchshark.dk is listed with a 17% increase in conversion rate, Frisør Sadon with 42% more calls and bookings plus 33% more time on site, and Shopsmukkere.nu with a 19% conversion lift.
Those case results came from different sites and should be treated as examples, not averages or promises for a new test.
The useful part is the controlled comparison.
Without a split test, a stronger opening line can look weaker simply because it ran during a slower week. Running both videos at the same time reduces that timing bias.
In CompleteGreet, turn on A/B Split Test for a single video bubble, keep Video A as the main video, upload Video B, and save. The feature is not available for Multivid bubbles.
Check the live page on mobile so the bubble does not cover checkout, chat, or form controls. Both video variants use the same bubble settings, so placement and timing belong in separate tests.
Choose one video winner before testing a second variable. If you change the video and placement together, you will not know which change caused the difference.
How to set up your first video A/B test
Open an existing single video bubble or create a new one, then prepare a second video for the comparison.
Keep both clips similar in length, framing, and production quality. Change one element, such as the first sentence. For example, compare an introduction with a direct question while keeping the same speaker and call to action.
In the bubble editor, switch on A/B Split Test. Video A is the current main video. Upload Video B, save the bubble, and leave the placement, trigger delay, buttons, and URL rules unchanged while the test runs.
Test on your actual phone, not just the browser inspector.
Use the same URL rules for both versions. If the bubble appears only on a pricing, contact, or checkout page, leave those rules unchanged during the test.
If you are testing the opening line, change only that line in Video B. Keep the bubble title and buttons the same so the result reflects the video change.
Fill out your own form after publishing. Submit it and check that the lead hits your inbox or CRM. Do this twice: once on desktop, once on mobile. A broken form can hide a strong video result.
The A/B dashboard uses bubble open rate to suggest a winner. It also reports estimated unique visitors, full views, button clicks, form submissions, and average watch time for each video.
CompleteGreet BUILD is $29 per month when billed yearly, or $36 month to month, for 5,000 unique visitors per month, 1 website, and 2 bubbles or videos. You can view pricing and plans to compare tiers.
Start with one bubble and two videos. Let the traffic and confidence checks pass before choosing the live video.
Testing video bubbles requires more than turning a feature on and off. Proper A/B testing isolates the video change and measures the result with statistical confidence.
The steps below match CompleteGreet’s current built in video test and analytics.
CompleteGreet waits for at least 1,000 estimated unique visitors per video, at least 3,000 total views, and 95% statistical confidence before suggesting a winner.
Run the test through at least one complete weekly traffic cycle, then keep it running until the dashboard thresholds are met. Do not stop as soon as one video moves ahead.
Use bubble open rate as the decision metric. Review full views, button clicks, form submissions, and average watch time to understand how people behaved after opening each video.
Before declaring a winner, confirm that the dashboard shows an even 50/50 split. Compare the live page on mobile and review Google Core Web Vitals guidance so a loading, interactivity, or layout difference is not mistaken for a video result.
Record any sale, launch, traffic campaign, or site change that happened during the test window. If it affected one part of the test more than the other, keep testing under normal conditions.
A few early visits can move the percentages sharply. Use the traffic and confidence checks in the dashboard instead of choosing a winner from an early lead.
If traffic is still low, the dashboard shows Keep testing. Choosing a winner sets that video as live and turns off the A/B split for the bubble.
Common questions
How long should I run an A/B test on a video bubble before calling a winner?
Cover at least one complete weekly traffic cycle, then wait until the dashboard has at least 1,000 estimated unique visitors per video, at least 3,000 total views, and 95% statistical confidence. A fixed number of days is not enough when traffic is low.
What metrics should I track when A/B testing video bubbles?
CompleteGreet uses bubble open rate to suggest a winner. The dashboard also shows estimated unique visitors, full views, button clicks, form submissions, and average watch time for Video A and Video B.
How do I set up two different video greetings for A/B testing?
Create or edit one single video bubble, switch on A/B Split Test, keep the current video as Video A, and upload Video B. CompleteGreet serves one version to each visitor automatically. Multivid bubbles do not support this test.
What is the sample size needed for a valid video bubble A/B test?
CompleteGreet waits for at least 1,000 estimated unique visitors per video and at least 3,000 total views. It suggests a winner only when the bubble open rate difference reaches 95% statistical confidence.
Can I run A/B tests on video bubble placement and timing?
The built in A/B test compares two videos while the bubble settings stay the same. To test placement or trigger timing, run a separate controlled test and change one setting at a time.
How much does it cost to run A/B tests with multiple video bubbles?
CompleteGreet BUILD is $29 per month when billed yearly, or $36 month to month, for 5,000 unique visitors per month, 1 website, and 2 bubbles or videos. The A/B feature uses Video A and Video B inside one single video bubble.
What is the fastest way to create a second video for A/B testing?
Record the second greeting on your phone or with the dashboard recorder. Keep the speaker, framing, and length similar, then change one element such as the opening line or call to action.
Do video bubbles slow down my site when running A/B tests?
CompleteGreet is designed to load after the main page content. Still test the live page on mobile and compare Core Web Vitals before and during the experiment instead of assuming every site will behave the same.
