14 {{ upvoteCount | shortNum }}
14 {{ upvoteCount | shortNum }}

How to split-test variations of creatives with POF

by inflameous
12 replies
Hello, I picked up 10 images and then did 7 variations of each using CrazyCTR, so totally I have 70 images, I want to start my campaign but I'm unsure how I should split-test them since I have 10 ad text variations, If I do 10 ad text variations to 70 ad images this will result in 700 creatives, which is very much and will burn my budget without even getting the amount of data for results. My questions is how should I split-test them.
#creatives #pof #splittest #variations
  • Profile picture of the author DavidLasry
    If you have low budget I would suggest you to test groups from your creatives.
    What I mean by that is don't test all of them at once. Take amount of ads you want to run and test them for a week. Then take another group and test them for another week. In this way you won't burn your budget very fast and if you have winners you can test different variations of them and get rid of the rest.
    Signature

    Sincerely,
    David.

    {{ DiscussionBoard.errors[8788346].message }}
    • Profile picture of the author inflameous
      My budget is about 500$, if needed I'll invest more.
      {{ DiscussionBoard.errors[8788402].message }}
  • Profile picture of the author ChrisBa
    Originally Posted by inflameous View Post

    Hello, I picked up 10 images and then did 7 variations of each using CrazyCTR, so totally I have 70 images, I want to start my campaign but I'm unsure how I should split-test them since I have 10 ad text variations, If I do 10 ad text variations to 70 ad images this will result in 700 creatives, which is very much and will burn my budget without even getting the amount of data for results. My questions is how should I split-test them.
    What you mentioned pretty much sums up how to do it. You could scale back to less creatives
    Signature
    {{ DiscussionBoard.errors[8788403].message }}
  • Profile picture of the author inflameous
    So should I split test by 100 creative's till I gather enough data for each?
    {{ DiscussionBoard.errors[8788458].message }}
  • Profile picture of the author Greedy
    Start with less text variations, text has much lower effect on the CTR, but does effect your conversion rates.

    Also I would focus on testing a lot of different images of people first, opposed to a few people with crazy boarders ect...

    Once you find a good person picture then start testing borders ect. Good Luck
    Signature
    {{ DiscussionBoard.errors[8789053].message }}
    • Profile picture of the author inflameous
      Thanks a lot Greedy, how many images should I aim to collect when starting a campaign?
      {{ DiscussionBoard.errors[8789098].message }}
    • Profile picture of the author sonixkid3000
      Originally Posted by Greedy View Post

      Start with less text variations, text has much lower effect on the CTR, but does effect your conversion rates.
      Please take this into account. Many people spend wayyy too much time on text variations and while they are helpful and do effect your ad performance, they won't make or break an ad. The image is the majority of what people see and focus on.

      What you wrote just about sums up how to split test. With that said, if you're looking to cut back costs a little bit, I typically tell people 2 text variations per image max.
      {{ DiscussionBoard.errors[8789138].message }}
      • Profile picture of the author inflameous
        Originally Posted by sonixkid3000 View Post

        Please take this into account. Many people spend wayyy too much time on text variations and while they are helpful and do effect your ad performance, they won't make or break an ad. The image is the majority of what people see and focus on.

        What you wrote just about sums up how to split test. With that said, if you're looking to cut back costs a little bit, I typically tell people 2 text variations per image max.
        Saw some case studies couple of days ago and they used about 5-10 text variations and I thought it's really important, thanks for the heads up.
        {{ DiscussionBoard.errors[8789149].message }}
  • Profile picture of the author DaveWarrior
    Test about 10 images at a time with 2 copies. Find the winners and replace the losers with new images. 10 copies is too much in the beginning because it will be too expensive as you said. Plus you can't really measure the effect of copy unless you are getting clicks so find the images that get you clicks first, then later on test your copies.
    {{ DiscussionBoard.errors[8789411].message }}
  • Profile picture of the author Greedy
    Thanks a lot Greedy, how many images should I aim to collect when starting a campaign?
    70 is great!

    Saw some case studies couple of days ago and they used about 5-10 text variations and I thought it's really important, thanks for the heads up.
    It is important, and not to be over looked. But it is just more important to have a good image at first on POF, since without a high CTR your campaign won't make it too far.
    Signature
    {{ DiscussionBoard.errors[8790299].message }}
    • Profile picture of the author sonixkid3000
      Originally Posted by Greedy View Post

      It is important, and not to be over looked. But it is just more important to have a good image at first on POF, since without a high CTR your campaign won't make it too far.
      ^^ What she said. lol

      It's not to say that's it something that people don't focus on but especially when starting out, focus on finding the right image. :p
      {{ DiscussionBoard.errors[8792722].message }}
  • Profile picture of the author Glynn Kosky
    If you had 70 images and 2 different AD copies, so 140 different ADS....how long should you run a test for to find a winner?
    {{ DiscussionBoard.errors[9046572].message }}

Trending Topics