Does Anyone Believe This Wild Claims? Software Says You Only Need 73 Visitors for An Accurate Test

11 replies
Conversion Prophet - Tracker That Predicts Conversion Rates


It seems this would be too good to be true. If this really worked you could charge $1,000 for the software. Nonetheless, it is interesting little case study. What do you guys things? Is anyone using this Taguchi software?

Originally posted in the cpa forum but this is probably a better place.
#accurate #claims #software #visitors #wild
  • Profile picture of the author Paul Hancox
    Hi Dave

    I haven't used it myself, but the software's "AIDA" feature seems to measure the length of time a visitor stays on a sales page, and then categorizes them into Attention, Interest, Desire and Action.

    I'm somewhat skeptical of Taguchi testing myself. On the other hand, the way they predict conversion rates based on visitor time on the site sounds quite useful (although calling it AIDA is somewhat of a "gimmick", methinks ... but a clever one).
    Signature
    PresellContent.com - How to sell without "selling"
    {{ DiscussionBoard.errors[2785471].message }}
    • Profile picture of the author ShaunAllen
      Really? Can you explain. It seems everyone has this down as a science. Especially the guy at split test accelerator.

      Originally Posted by Paul Hancox View Post

      Hi Dave

      I haven't used it myself, but the software's "AIDA" feature seems to measure the length of time a visitor stays on a sales page, and then categorizes them into Attention, Interest, Desire and Action.

      I'm somewhat skeptical of Taguchi testing myself. On the other hand, the way they predict conversion rates based on visitor time on the site sounds quite useful (although calling it AIDA is somewhat of a "gimmick", methinks ... but a clever one).
      Signature

      {{ DiscussionBoard.errors[2786567].message }}
      • Profile picture of the author Paul Hancox
        Hi Dave

        Originally Posted by DaveJ13 View Post

        Really? Can you explain. It seems everyone has this down as a science. Especially the guy at split test accelerator.
        Well, it IS a science, no doubt about that. The important question for copywriters is whether its suitable for testing or not.

        Taguchi is great for testing the manufacture of CARS, which is what it was designed for. Each component of the manufacturing operates within strict, predictable limits, which is why it works in that context.

        However, sales copy doesn't work in quite the same mechanical way as car components.

        For example, let's say we have:

        Headline 1 / Headline 2 / Headline 3

        Copy 1 / Copy 2 / Copy 3

        Offer 1 / Offer 2 / Offer 3

        Due to psychology, let's say a combination of Headline 1, Copy 2 and Offer 3 is going to work best.

        Taguchi testing does NOT test all combinations, but only a certain number of them, and then predicts based on those combinations.

        However, because it is a MATHEMATICAL tool but doesn't understand copywriting, psychology etc... Taguchi testing may not recognize (or even test) the combination that is going to win!

        That's why I'm skeptical of Taguchi testing. Straightforward multivariate testing tests ALL combinations, and puts each to the vote of the visitor's wallets and purses.

        On the other hand, I do think there is quite a lot of validity to the idea of using length of time on page as a predictor of conversions.
        Signature
        PresellContent.com - How to sell without "selling"
        {{ DiscussionBoard.errors[2787289].message }}
        • Profile picture of the author RickDuris
          I have not used this product. But with traditional Taguchi testing, the answer depends upon the number of variations you are testing at one time.

          The thing about Taguchi is, how you test and what you test is more important than the software. You must be trained to appreciate the strategy behind Taguchi.

          It's not so simple, even though results can be optimized in a fraction of the time.

          - Rick Duris

          PS: What I find is marketers have trouble finding the time to do a/b split testing, let alone Taguchi.
          Signature
          {{ DiscussionBoard.errors[2787450].message }}
    • Profile picture of the author Matt MacPherson
      Originally Posted by Paul Hancox View Post

      Hi Dave

      I haven't used it myself, but the software's "AIDA" feature seems to measure the length of time a visitor stays on a sales page, and then categorizes them into Attention, Interest, Desire and Action.

      I'm somewhat skeptical of Taguchi testing myself. On the other hand, the way they predict conversion rates based on visitor time on the site sounds quite useful (although calling it AIDA is somewhat of a "gimmick", methinks ... but a clever one).
      Do you know of any programs that use scrolling to predict conversion. Could be really useful for anything above the fold.

      Matt
      {{ DiscussionBoard.errors[2788046].message }}
  • Profile picture of the author maximus242
    Its doesn't make statistical sense, a statistician (ive known a few) would say its entirely bull****, it doesn't have a large enough group of data to counteract for the possibility of random upswings or downswings in conversion
    Signature

    xResponsive Advertising Agency | Direct Marketing | Online Advertising | Create Breakthrough Campaigns for Your Business http://xresponsive.com

    {{ DiscussionBoard.errors[2786625].message }}
    • Profile picture of the author ShaunAllen
      How does everyone test then?
      Signature

      {{ DiscussionBoard.errors[2786700].message }}
  • Profile picture of the author maximus242
    You use larger data than 50 or 70, the results can be inaccurate. Ive had test results at 100 where it looked like one was the winner and at 200 it completely reversed
    Signature

    xResponsive Advertising Agency | Direct Marketing | Online Advertising | Create Breakthrough Campaigns for Your Business http://xresponsive.com

    {{ DiscussionBoard.errors[2787037].message }}
  • Profile picture of the author nemock
    Merely testing with 73 visitors probably beats the amount of testing done by most people. Even with such a small sample, it might give enough cues about differences between versions to get results.

    ...and in the end, results are results. ;-)
    {{ DiscussionBoard.errors[2787789].message }}

Trending Topics