7 {{ upvoteCount | shortNum }}
7 {{ upvoteCount | shortNum }}

Website Optimizer Split-Test Question!

by SeanyG
6 replies
Hey Website Optimizer pros!

I'm just setting up the tracking scripts for my different multivariant experiments and ran into an issue.

I want to split test the headlines at the top of my salespage (under the sales video).

I also want to test what is the better placement for the headline, above the sales video or below (where the headline is now).

So I added the section tracking script to the headline. It works great (validated).

But then when I go to create the experiment where the original headline is below the video and a the split test (new option) is above, it gives me errors because the scripts overlap. Obviously the headline is included in both scripts and optimizer doesn't like this...

Is there any way to split test an image (headline) while also split testing its position on the page?

Thanks!

~Sean
#optimizer #question #splittest #website
  • Profile picture of the author Chris Thompson
    I don't know the answer to your question, but what I do is simply run A/B experiments and just keep trying new combinations. So test your headline first and then test position of headline after.

    Mutlivariate = VERY LONG TESTING CYCLE unless you have a crapload of traffic.
    {{ DiscussionBoard.errors[3679560].message }}
    • Profile picture of the author SeanyG
      Originally Posted by Chris Thompson View Post

      I don't know the answer to your question, but what I do is simply run A/B experiments and just keep trying new combinations. So test your headline first and then test position of headline after.

      Mutlivariate = VERY LONG TESTING CYCLE unless you have a crapload of traffic.
      Hey Chris!

      Great advice. Thanks for the reply.

      So you just do 1 a/b test at a time? And you end up getting results faster?

      Thanks!

      ~Sean
      Signature
      FREE Video DJ Lessons to help you learn
      >>>How To DJ
      {{ DiscussionBoard.errors[3682131].message }}
      • Profile picture of the author Trivum
        Originally Posted by SeanyG View Post


        So you just do 1 a/b test at a time? And you end up getting results faster?
        I don't know about "faster," but you get more reliable info. The way to split test is to always only change ONE variable. If you change more than one variable, you have no idea what accounted for the results you see.
        {{ DiscussionBoard.errors[3682181].message }}
        • Profile picture of the author SeanyG
          Originally Posted by Trivum View Post

          I don't know about "faster," but you get more reliable info. The way to split test is to always only change ONE variable. If you change more than one variable, you have no idea what accounted for the results you see.
          Great point! Thanks Trivum.

          That being said I have LOTS of things I want to split test.
          - Layout
          - Headlines
          - Images
          - Sales videos
          - Order buttons
          - backgrounds
          - Many more!

          With so many different variables I want to test is it still better to do it A / B?

          I figured it would be better to just set up all of the variable is a multivariate experiment and just let them all run at the same time.
          Signature
          FREE Video DJ Lessons to help you learn
          >>>How To DJ
          {{ DiscussionBoard.errors[3682195].message }}
          • Profile picture of the author Trivum
            Originally Posted by SeanyG View Post

            Great point! Thanks Trivum.

            That being said I have LOTS of things I want to split test.
            - Layout
            - Headlines
            - Images
            - Sales videos
            - Order buttons
            - backgrounds
            - Many more!

            With so many different variables I want to test is it still better to do it A / B?

            I figured it would be better to just set up all of the variable is a multivariate experiment and just let them all run at the same time.
            The logic still holds no matter how many variables you have. There's no way you'll be able to tell what's "best" for each of them unless you parse each one out and test it. ... More work, yes, but there's no way around it if you want the "truest" results possible.
            {{ DiscussionBoard.errors[3682234].message }}
            • Profile picture of the author SeanyG
              Originally Posted by Trivum View Post

              The logic still holds no matter how many variables you have. There's no way you'll be able to tell what's "best" for each of them unless you parse each one out and test it. ... More work, yes, but there's no way around it if you want the "truest" results possible.
              Thanks Trivum! Not more work but more time.

              But if it provides more accurate results I'll use A/B!
              Signature
              FREE Video DJ Lessons to help you learn
              >>>How To DJ
              {{ DiscussionBoard.errors[3691833].message }}

Trending Topics