• Guest - Earn a FREE TubeBuddy Upgrade for being active on the forums! Click Here to learn how you earn free upgrades for TubeBuddy!
  • Guest - TubeBuddy has a discord! Click Here to join in the conversation!

TubeBuddy Suggestion Expanded A/B testing using 2 variations

Beanie Draws

Mythical Poster
2,883
27
www.youtube.com
Subscriber Goal
30000
So I'm working on re-doing a lot of my old thumbnails (as always) and I have 2 new thumbs that are somewhat differant in theme, and I've come to realise I can only A/B test one at a time.

Is it possible to expand the A/B testing option to have, say, 3 alternatives max, as an option?

Say I want to test 2 new thumbnails, and I want to see how they perform compared to each other, and compared to the original.

The process might be
New A - Old B - New C - Old B - New A - Old B - New C - Old B - New A - Old B - New C - Old B - (test new thumbnail, then compare it to the old, test second new thumbnail, compare to the old, then you can see how the 2 new thumbnails compare against each other, compared to the original)
or
New A - New B - Old C - New A - New B - Old C - New A - New B - Old C - New A - New B - Old C - (basically the same, but differant sorting, new thumbnail, second day another new thumbnail, third day revert back to the original. theoretically should get the same type of data, with just a differant method of shuffling)


If the test runs long enough, you can still theoretically get enough data to test more than one option. I know, I like to make things complicated :joy: t'is but an idea/suggestion :)
 

TubeBuddy

Legendary Poster
12,445
33
www.tubebuddy.com
Subscriber Goal
5000
So I'm working on re-doing a lot of my old thumbnails (as always) and I have 2 new thumbs that are somewhat differant in theme, and I've come to realise I can only A/B test one at a time.

Is it possible to expand the A/B testing option to have, say, 3 alternatives max, as an option?

Say I want to test 2 new thumbnails, and I want to see how they perform compared to each other, and compared to the original.

The process might be
New A - Old B - New C - Old B - New A - Old B - New C - Old B - New A - Old B - New C - Old B - (test new thumbnail, then compare it to the old, test second new thumbnail, compare to the old, then you can see how the 2 new thumbnails compare against each other, compared to the original)
or
New A - New B - Old C - New A - New B - Old C - New A - New B - Old C - New A - New B - Old C - (basically the same, but differant sorting, new thumbnail, second day another new thumbnail, third day revert back to the original. theoretically should get the same type of data, with just a differant method of shuffling)


If the test runs long enough, you can still theoretically get enough data to test more than one option. I know, I like to make things complicated :joy: t'is but an idea/suggestion :)

I guess the hard part is, knowing if the old one will for sure do better than the new one. With 3 it becomes more complicated to read the data exactly. I could argue you could do the inital A/B test find out which one does better, than do the new one based on the results of A/B test #1 I'll send this to the team. That's just my initial thoughts :)
 
  • Like
Reactions: Beanie Draws

skrach

New Member
22
7
Subscriber Goal
5000
This is an area where I'd personally recommend an external tool. Even doing this "live" on YouTube might not give you great comparative results in a short period of time.

What would accomplish your goal though, is Google ads via clicking the promotion button in studio.

You can set up a campaign where all results land on your video, but have three or more different thumbnails set up as different ads within the campaign. You should be able to run them all simultaneously and under the same budget to see which thumbs get a higher CTR and view through rate.

I realize your request is meant for Tubebuddy, but the Google ads option is available now. If you don't want the results of the "promotions" to affect your videos - at least according to YouTube - you can either 1) set the videos to unlisted for the duration of the testing or 2) if you typically have a large volume of viewers (at 26k subscribers, you should) set a negligible overall budget.

While this option is NOT free, you'd be paying for the value of the data. And since Google ads doesn't require a minimum spend, you can decide what that is.

<- NOT a Google ads employee, just a Youtuber that uses their services to jumpstart channel growth.