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

How and what do you test

by agmccall
6 replies
Hello All

I hear all the time about testing, testing, testing. Split tests and everything that goes with it but I am confused so I am hoping for a bit of help. Here is a banner. I am not in this niche, the banner is part of some banner template software that I have. I just made a few edits to make it complete.



So I go ahead with a campaign what do I test, where do I start, what do I change

I know the question is a little vague but I am really trying to wrap my head around this and can not seem to grasp the concept.

Thank You All

al
  • Profile picture of the author EugeneFarber
    The short answer is: everything!

    The slightest changes can have the most profound effects. For example, changing the arrow from blue to red might be something you can test. Sounds trivial but it can really change click-throughs rates.

    Test the images, the headings, the copy...everything!

    Also, don't forget about the landing page that the ad leads to.

    I tested a squeeze page where I kept all copy the same and only changed the background image. The results showed one background image converting at over 300% higher than the control.

    It's amazing how different elements can have the biggest effects.

    So again...test everything!
    {{ DiscussionBoard.errors[7821539].message }}
    • Profile picture of the author agmccall
      Seems overwhelming and costly, although, probably worth the cost.

      So after I design my banner and squeeze page, just start making subtle changes and see what works

      al
      Signature

      "Opportunity is missed by most people because it is dressed in overalls and looks like work." Thomas Edison

      {{ DiscussionBoard.errors[7821556].message }}
      • Profile picture of the author WillR
        Originally Posted by agmccall View Post

        Seems overwhelming and costly, although, probably worth the cost.

        So after I design my banner and squeeze page, just start making subtle changes and see what works

        al
        It is overwhelming, definitely. But if you are not prepared to test and track everything then you won't last long in paid advertising and you will lose a lot of money. In fact most media buyers will tell you to expect losing money for a while. It's the optimizing and testing of campaigns that usually finds you what works. 9 out of 10 campaigns will probably make you no money but you need to go through them to find the 1 that does work.

        If the banner in your first post is a real banner you are using, I'm confused as to the benefit or reason for anyone to click on it? It doesn't make a whole lot of sense to me. I would be finding a much stronger headline or creating more curiosity.
        Signature

        .


        .

        {{ DiscussionBoard.errors[7822944].message }}
  • Profile picture of the author GotLiveChat
    What is a good length of time to evaluate different ad types, before switching? Are we talking days, weeks or months?
    {{ DiscussionBoard.errors[7821587].message }}
  • Profile picture of the author Stuart Walker
    The texts, the images, the colours, the call to action, the placement of the banner.

    Seems overwhelming and costly
    Ok well don't bother then and just stick with whatever conversion rate you get and never find out if it can be improved....

    So after I design my banner and squeeze page, just start making subtle changes and see what works
    Yes basically. You can set up a Google content experiement so that different visitors see different version of the page and see what converts the best.
    Signature
    ATTENTION WARRIOR: INCREDIBLE NEW POST AT NICHEHACKS.COM
    {{ DiscussionBoard.errors[7821619].message }}
  • Profile picture of the author EugeneFarber
    Overwhelming...yea...at first it definitely might be. But costly...not so much (at least in terms of dollars, maybe not time).

    You can use Google Analytics (with events and content experiments) to track links and conversions. It's all free.

    You can also set the statistical confidence level you want to reach within the Google experiment (I believe the default is 95% confidence). However, I would run it a bit longer past that point.
    {{ DiscussionBoard.errors[7822101].message }}

Trending Topics