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

Testing is everything... Discuss the best method

by Davids22
6 replies
We all know that testing is one of the most important parts of a business, but how do you know what to test, and how, or when?

Here are 100 conversion case studies by KissMetrics (not sure if shared before):
https://blog.kissmetrics.com/100-con...-case-studies/

Some of the changes seem very small, but have a dramatic impact, other changes are extremely dramatic, and cause less of an impact on conversions.

So my question is as follows... how do you test your ads/landing pages?

My strategy:
Send traffic to 2-3 hugely different landing page layouts, wording, and angles.

Take the best converting one, add a completely different one, and add a third, a tweak of the 'control' by changing the headline, the colors, or something else (but only 1 at a time).

Then move on from there, forever, and ever.

Same method for my ads.

However, I base all of my decisions on a statistically significant result, using a calculator online: AB split test graphical Bayesian calculator

This allows me to remove all emotions from the equation and actually work on data driven impacts, which is what matters in any business.

What's your process?

What do you test first?
#ab split #multivariate #statistical significance #testing
  • Profile picture of the author goozleology
    When I create and run ads, I create four different ads and run them until I get at least 1000 clicks. Once I have a winner, I let that ad run 75% of the time and a new ad 25% till I get 5000 clicks total and I keep the winner.

    I also do something similar with landing pages where I run 75% of my traffic to the winner and 25% to a challenger. If the challenger beats the champ then I run it 75% of the time and create a new landing page for the other 25%. I do this because I because a bad landing page can hurt more than the benefits of a good landing page.

    I hope the makes sense.
    {{ DiscussionBoard.errors[9608247].message }}
    • Profile picture of the author Davids22
      Originally Posted by goozleology View Post

      When I create and run ads, I create four different ads and run them until I get at least 1000 clicks. Once I have a winner, I let that ad run 75% of the time and a new ad 25% till I get 5000 clicks total and I keep the winner.

      I also do something similar with landing pages where I run 75% of my traffic to the winner and 25% to a challenger. If the challenger beats the champ then I run it 75% of the time and create a new landing page for the other 25%. I do this because I because a bad landing page can hurt more than the benefits of a good landing page.

      I hope the makes sense.
      Yeah, that makes total sense, and I completely agree.

      Though I find you can statistically optimize before the total of 1,000 clicks, especially if you can see impressions on the ads.

      I should have mentioned my %'s, but it's somewhat similar to your technique.

      Thanks for the share
      Signature
      GorillaBuild.com - Easy, drag & drop funnel builder.
      {{ DiscussionBoard.errors[9608692].message }}
  • Profile picture of the author Sarevok
    It's easy for me.

    I basically have two "major" segments in my email list. (Leads, and Buyers).

    I always test stuff to my leads list first... And if it works?

    I know it's going to CRUSH when I mail to my buyer's lists.



    (There's more than one way to slice the cake right? Well, this one works really easy, and helps protect my buyers by only offering them stuff that's proven to rock).
    Signature
    {{ DiscussionBoard.errors[9608260].message }}
    • Profile picture of the author Davids22
      I'm not much of a list builder, though I am moving in that direction.

      If you're just promoting affiliate stuff and not your own products it could be somewhat hard to build a buyers list (unless you have a good relationship with the product owner).

      Do you also segment your lists by age groups/gender/etc? I could imagine promoting things that fall into those categories as an 'ooo look what I/my wife found' e-mail.

      Do you think this would work?


      Originally Posted by Sarevok View Post

      It's easy for me.

      I basically have two "major" segments in my email list. (Leads, and Buyers).

      I always test stuff to my leads list first... And if it works?

      I know it's going to CRUSH when I mail to my buyer's lists.



      (There's more than one way to slice the cake right? Well, this one works really easy, and helps protect my buyers by only offering them stuff that's proven to rock).
      Signature
      GorillaBuild.com - Easy, drag & drop funnel builder.
      {{ DiscussionBoard.errors[9608698].message }}
  • Profile picture of the author DURABLEOILCOM
    You can test traffic with Google Analytics.
    Signature


    Ready to Be The Bo$$!?!

    Start Your Own Business Today!
    {{ DiscussionBoard.errors[9608714].message }}
    • Profile picture of the author Davids22
      Originally Posted by DURABLEOILCOM View Post

      You can test traffic with Google Analytics.
      ???

      That's for tracking, not testing.

      I opened this thread to discuss the methods you use to test certain elements of any given campaign you are running, and I was hoping for a more in-depth, analytical answer.

      What data does Google Analytics provide you that shows you or gives you ideas on what to test?

      Cheers
      Signature
      GorillaBuild.com - Easy, drag & drop funnel builder.
      {{ DiscussionBoard.errors[9608757].message }}

Trending Topics