13 {{ upvoteCount | shortNum }}
13 {{ upvoteCount | shortNum }}

Incredibly Powerful Multivariate Split Testing Software - Free Test To 10 Lucky Warriors

by TimRobinson
11 replies
Hey All,
Been working on some huge new features for the Evolutionary Split Tester over the past few months. Seeking 10 Beta testers who want to get their hands dirty with (IMO) one of the best Multivariate Split Testers out there.

I'm Planning on spending a lot of time over the next two weeks fixing every single bug adding all the ideas to develop the most incredibly powerful, fantastic, groundbreaking version ever, then raising the price.

To get your chance to be a tester reply here or send me a PM with a short reason for why you'd make a great tester.

and yes if you really feel like purchasing it right now (and to everyone who has already purchased) you'll get free updates and patches for life (so if you want to save yourself some money, get in early)

Cheers, Tim
#beta #evolutionary split tester #multivariate #software #split #split testing #split testing software #testers #testing software
  • Profile picture of the author YiKeS
    G`day Tim,

    I`ll give the software a run mate, add me
    to Skype `n hook up, I`m a few hundred
    K`s north of you to btw

    Christopher J.
    Signature
    AUSSIE MADE
    WSO --> HOMEPAGER <--- WSO
    {{ DiscussionBoard.errors[242073].message }}
  • Profile picture of the author TimothyW
    Hi Tim...

    Looks good. I wouldn't be a candidate for the testing - but I have a q about the current version. Is there a way to set the 'threshold' of the testing? I haven't looked at the 12-minute video yet, so maybe this is answered there.

    Basically I want to know if it's possible to set how many times a given variation must 'lose' before it is dropped from the testing?

    Here's a simple example -- let's say you just want to test 5 diff. headlines. Does it test all 5 *evenly* until you tell it to stop? Or does it automatically STOP testing the losing headline when IT figures out which one is the losing one? And if so, how many times does it test (try) all 5 headlines before it starts NOT testing the losing one?

    Reason I'm asking this is, if it drops a headline, JUST because that headline 'lost' after just trying that headline *ONCE*, then it's not gathering enough 'evidence' to make convincingly accurate decisions. See what I mean?

    The user should be able to set the level or threshold of the decision making process, if possible. Like, 'give these 5 headlines 10 chances each before deciding to eliminate the bottom one (the least successful one), then test the remaining 4 headlines 10 more times each, etc.

    See what I'm getting at?

    It's about how many chances a certain variation is given before e decision is made. The more chances given, the more 'accurate' (meaningful) the decision is.

    Can those thresholds be set by the user with your software?

    Thanks.

    -- TW
    Signature
    {{ DiscussionBoard.errors[242286].message }}
    • Profile picture of the author TimRobinson
      Instead of simply showing every headline/body paragraph etc evenly (which would cause you to lose sales as a crappy headline may show for a while before being dropped out of rotation) what it does is use a specially designed system to show each element more or less often as it performs better or worse.

      For example if you have 3 headlines:

      Headline A Converts at 20%
      Headline B Converts at 3%
      Headline C Converts at 2%

      The system realizes that headline A is far out performing the other 2 headlines and as such it displays it far more often than the others.

      Headline A will have a 80% chance of showing
      Headline B will have a 12% chance of showing
      Headline C will have a 8% chance of showing

      This is all done in real time as visitors come along to create a constantly evolving sales page that changes itself to suit what people desire automatically.

      Now I found a problem with this. In the beginning I noticed if one headline got just one click through while the others got none the others wouldn't show. So I fixed the problem with a simple solution:

      You are able to set the minimum clickthrough rate for a headline so it keeps showing even if it has a bad click through rate until a certain amount of impressions is reached.

      So if you set a minimum 10% click through rate the headlines above would look like so:

      Headline A: 20%
      Headline B: 10%
      Headline C: 10%

      And they would have the following chance of showing:

      Headline A: 50%
      Headline B: 25%
      Headline C: 25%

      This is so that a good headline has a fair chance of being shown to a lot of visitors to determine what is really the best headline.

      All this maths is done in real time by the SplitTester so it is constantly updating and changing your sales page to make as many sales for you as possible.

      If a headline performs poorly it will be shown much less often so it won't lose you sales.

      I don't have a system in place to make tests drop off if they fall below a certain percentage although that sounds like an excellent idea TW! (will get around to implementing it this patch )

      Along with this new patch I'll be recreating about 10 tutorial videos because the current ones are a bit old :p.
      Signature
      ** Free Visual Split Testing and Heatmapping software that will rapidly increase your copywriting results (and profits) by giving you access to successful and failed tests hundreds of others have run **
      {{ DiscussionBoard.errors[242345].message }}
      • Profile picture of the author Reeveso
        Sounds pretty cool...can you explain how it's different from Google's free multivariate split-tester?

        I'm not trying to make you look bad or doubting you or anything, I'm just wondering. Also maybe it'll give you some ideas on how to make it different so people have a reason to buy yours instead of get it free from Google.

        I'd be willing to test it out. I have a landing page I'm going to be testing in the next week or so that I could use it on and give you some feedback.
        {{ DiscussionBoard.errors[242392].message }}
        • Profile picture of the author TimRobinson
          Just posted a new entry on my blog addressing this exact issue , so heres the lowdown:


          A Lot of people have been asking me "Why would I use the Evolutionary Split Tester when Google website optimizer is free and appears to do the same thing?".

          We'll heres the lowdown: Google Website Optimizer is COSTING you a LOT more money than you think!

          Don't believe me? I'll let the numbers tell you the story.

          Lets say you've just launched a new product: a $27 Ebook. To test its conversion rate you want to send a flood of traffic to it immediately, lets say 1000 visitors a day for a week (7 days).

          For the sake of simplicity I'll say you're only testing on variable: the headline (although I do realize both testers can test multiple variables at once, this is just so we can see how the maths works), and you decide to test 4 different headlines.

          Headline A Converts at 3%
          Headline B Converts at 1%
          Headline C Converts at 0.5%
          Headline D Converts at 0.5%

          Now on the first day both split testers have no data, so they show each headline to the same amount of visitors, which would be 250 (1000 / 4).

          Thus you make the following amounts in sales:

          Headline A = 3% * 250 = 7.5 * $27 = $202.50
          Headline B = 1% * 250 = 2.5 * $27 = $67.50
          Headline C = 0.5% * 250 = 1.25 * $27 = $33.75
          Headline D = 0.5% * 250 = 1.25 * $27 = $33.75

          Total = $337.50

          Now what the Evolutionary Split Tester and Google Website Optimizer do differently at this point, is the Evolutionary Split Tester automatically adjusts your sales page to show the better performing tests more often, so Headline A would show 3 x as much as headline B and 6 x as much as headlines C and D (as it converts that much better). While Google will still show all of them evenly. Lets see how this stacks up in day 2's stats.

          The Evolutionary Splittester will show each headline this amount of times (based on it's conversion rate):

          Headline A = 60%
          Headline B = 20%
          Headline C = 10%
          Headline D = 10%

          Google Analytics:

          Headline A = 3% * 250 = 7.5 * $27 = $202.50
          Headline B = 1% * 250 = 2.5 * $27 = $67.50
          Headline C = 0.5% * 250 = 1.25 * $27 = $33.75
          Headline D = 0.5% * 250 = 1.25 * $27 = $33.75

          Total = $337.50

          Evolutionary Splittester:

          Headline A = 3% * 600 = 18 * $27 = $486
          Headline B = 1% * 200 = 2 * $27 = $54
          Headline C = 0.5% * 100 = 0.5 * $27 = $13.50
          Headline D = 0.5% * 100 = 0.5 * $27 = $13.50

          Total = $567

          In Just 1 Day Google has LOST you $229.50! Thats more than 4 times the cost of the Evolutionary Split Tester.

          Over the course of the week if you just keep them both running You will make:

          Google = $337.50 * 7 = $2,362.50
          Evolutionary Split Tester = $337.50 + $567 * 6 = $337.50 + $3402 = $3739.50

          Difference = $1377

          By Using Google for that week you've lost over $1377 in sales!

          Of course after this point most people will go into their stats, realize one headline is vastly outperforming the others and delete the under performing ones.

          However in just 1 week (most people I know won't touch their sales page for a week to give it time to fully test itself out) you've already lost over $1,300 in sales, over 10 x as much as the Evolutionary Split Tester costs!

          In case you're wondering: yes the Evolutionary Split Tester does adjust all its results in real time, not at the end of the day, thus even on the first day it would be outperforming google in sales.

          Questions? Comments?

          Cheers, Tim
          Signature
          ** Free Visual Split Testing and Heatmapping software that will rapidly increase your copywriting results (and profits) by giving you access to successful and failed tests hundreds of others have run **
          {{ DiscussionBoard.errors[243822].message }}
          • Profile picture of the author Reeveso
            All I have to say is...

            Great job at differentiating the product and making it worth the money
            {{ DiscussionBoard.errors[244016].message }}
            • Profile picture of the author YiKeS
              G`day,

              Tim, still waiting to hear from you, are
              you scared I`ll break it mate? :rolleyes: lol

              Christopher J.
              Signature
              AUSSIE MADE
              WSO --> HOMEPAGER <--- WSO
              {{ DiscussionBoard.errors[244171].message }}
  • Profile picture of the author getsmartt
    I'm interested in beta testing this, is it compatible with ads served from OpenX?
    Signature

    Was mich nicht umbringt, macht mich stärker

    {{ DiscussionBoard.errors[244197].message }}
    • Profile picture of the author TimRobinson
      @Christopher J: Still ironing out a few new bugs/features, will get back to everyone who has signed up shortly

      @getsmartt Currently it's only built for a single sales page where there is one goal in mind, not multiple possible goals. So it wouldn't work if you have a few ads and want to track clicks to all of them.

      If you have just one advertising spot and you wish to rotate between ads then track if the user clicked through to the buy page that will work, however if you'd like to have multiple ads on the page and track click throughs for all at once it won't.

      Will look into implementing that in the near future.
      Signature
      ** Free Visual Split Testing and Heatmapping software that will rapidly increase your copywriting results (and profits) by giving you access to successful and failed tests hundreds of others have run **
      {{ DiscussionBoard.errors[244402].message }}
      • Profile picture of the author YiKeS
        Originally Posted by TimRobinson View Post

        @Christopher J: Still ironing out a few new
        bugs/features, will get back to everyone who has signed up shortly

        @getsmartt Currently it's only built for a single sales page where there
        is one goal in mind, not multiple possible goals. So it wouldn't work if you
        have a few ads and want to track clicks to all of them.

        If you have just one advertising spot and you wish to rotate between
        ads then track if the user clicked through to the buy page that will work,
        however if you'd like to have multiple ads on the page and track click
        throughs for all at once it won't.

        Will look into implementing that in the near future.

        G`day Tim,

        no worries mate ... looking forward to it ... one thing about the A.I.
        automatically adjusting "show" percentages ... is it variable on the time
        taken to do this? You say it will optimize after 24 hours ... can this be
        extended for low traffic pages? Or can it be set by the # of visitors
        before optimization takes place? I`d hate the software to make a
        desicion after say 10 people .. lol

        Probably should just watch the vid aye!! :rolleyes:

        Christopher J.
        Signature
        AUSSIE MADE
        WSO --> HOMEPAGER <--- WSO
        {{ DiscussionBoard.errors[244502].message }}

Trending Topics