Can you run a multivariate test for a soft drink? 

Yesterday, I came across a product which was obviously running a multivariate test for their product.  But, this wasn’t an online business.  This was a soft drink company.

For a number of years now, I’ve noticed some number and letters under the bottle cap.  Before that company was putting numbers and letters under the cap, they originally had a little message that said something like “free 12oz drink”.  If you won something, you could just take it to the store and get a free drink.  It was a little trick they played.  They did whatever they could do to get you to look under that bottle cap.  For a good reason…  You might have won a free drink!  But lately, I haven’t seen that.  I’m not saying they no longer give away free drinks, but instead, they just show some numbers and letters.

So how the heck is that a multivariate test?

Well, if you think about it, it’s no different than the new feature in MuVar which converts an IP address into an address separated by dashes to make it look more like a coupon code.  You can use that feature with any physical product you sell online through a website.

Of course, MuVar is a bit different because MuVar is using an IP address instead of a pre-made code which is placed on the product itself.  If you were printing your own bar codes, and manufacturing your own physical products instead of using a service, then I don’t see why you couldn’t set it up the same way as the soft drink company I noticed.

So, back to the soft drink.

You see the code under the bottle cap.  Then what?  Well, there are instructions on the bottle label.  It tells you to go to a website where you can enter your code to win rewards.  You sign up, and by the time you’ve entered your code, the company now has all of your contact information as well as the code of the bottle which was presumable purchased.  If they’re using the code for multivariate testing (and I’m only assuming they are, since I don’t work with them) then they now know which bottle label was profitable.

Interesting?

I thought so…

That reminds me.  Another soft drink company likes using quotes on their bottle caps.  While that does keep some people entertained (for like 5 minutes), it probably doesn’t help the company one bit in increasing their profits.

- Aaron Brandon

I was just taking a look at one of my MuVar tests and realized that a few variable versions had a shocking confidence rate.  A nice 99% confidence rate!  But what was even more shocking was that the two variable versions were completely different…

Not different as in the same variable with a slightly modified version…this test was pretty much like taking one version, then throwing the other one out the window.  This variable version only rarely gets tested by anyone… or even by the “guru” types.  In fact, I’ve never heard one of those so called “guru’s” ever mention this type of test variable.

It’s VERY powerful.

If you add this one variable version to some of your own tests, you’re pretty much guaranteed to make a shocking discovery.

I would like to share these test results with my readers.  However, because they have a certain value to them, I’d like to keep this private rather than using just a regular blog entry.

So this is what I’m going to do… I’m going to make a private, password protected blog entry on Thursday morning (Feb 21) including these test results.

I’ll be sending out the password to my blog announcement list a few hours prior to posting the blog entry.  Of course, that means if you aren’t currently on my blog announcement list, I won’t be sending you the password.

So, to ensure you get the password, you must fill out your first name and primary e-mail address in the form on the top right of my blog (just under my picture there), and click on the subscribe button.  Once you’ve done that, make sure you check your inbox for a confirmation e-mail so I know I have your permission to send you e-mail. (remember to check your spam folders if you can’t find it)

But, don’t try using those fake, temporary e-mail accounts just for getting the password.  Some readers have tried to scam me in the past in that sort of way.  That doesn’t fly with me.  I ban folks who do tricky things like that.  Please use a REAL e-mail account.

Alright, go sign up now before you forget… and I’ll be sending you an e-mail soon!

Thanks!

- Aaron Brandon

You’ve got to take a look at this:

http://www.conversiondoctor.com/conversion-blog/2008/glyphius-after-181-test-rounds-how-does-it-really-stack-up/

- Aaron Brandon

I’ve decided to re-release my report on how to run a multivariate test using MuVar and Aweber (or any other autoresponder).

This time, it’s for only 24 hours.

If you need a bit more of a description, you can read the previous special offer I wrote, here: http://www.aaronbrandon.com/?p=190

This is where you purchase the report: http://www.aaronbrandon.com/special.html

After ordering, you’ll be sent to a thankyou page. From there, you’ll simply enter your name and e-mail address in the form provided, hit the submit button, and you’ll be sent to the page with the report. You will also receive an e-mail from me with the URL to access the report.

PLEASE NOTE THAT THIS OFFER ENDS ON THURSDAY AT 12 NOON SHARP (PST).

I DON’T WANT ANY OF MY READERS TO HUFF AND PUFF BECAUSE THEY MISSED THIS SPECIAL. SO IF THIS IS OF ANY INTEREST TO YOU, GRAB IT WHILE YOU CAN!

- Aaron Brandon

Like the headline? I thought it was kind of catchy.

I wanted to write a quick post about “nothing” here.  And, before you run off thinking that I’ve completely lost my mind, I need to inform you that…

I’m referring to copywriting!!!

You see, when I run multivariable tests, I include a number of variable versions. In those tests, I also include a version called “nothing”.  That version is in fact, blank.

There’s always the possibility that your variable doesn’t belong on the page at all… it could even depend on who your target market is.

I recently found out that my opt-in form (the one on the top right corner of this page…unless you’re reading this from an RSS reader, in that case you’ll have to come to my page and check it out) doesn’t perform as well with a headline. At least, not the headlines I used (could they really be that bad?).

Recent data taught me the following:

Version: 1 - Subscribe Now!
Conversion Ratio: 0.82%

Version: 2 - Nothing
Conversion Ratio: 1.64%

Version: 3 - Subscribe!
Conversion Ratio: 0.57%

The opt-in form which showed versions 1 and 3 had less opt-ins than “nothing” did. “Nothing” actually had 4X more opt-ins than the other two…

Interesting?

I thought so…

- Aaron Brandon

After posting the initial test results from the multivariate test I’ve been running on my opt-in form , some readers expressed interest in learning how to run their own similar multivariate opt-in tests. I know many of my readers already own the software MuVar, or MuVar 2007 which is the software I’m using to run the test. However, most of those owners only know how to test a sales page.

I also know some of my readers also use Aweber as their autoresponder service. Though Aweber allows you to split test or A/B test your web forms, the amount you can customize is very minimal. So, to customize further, you need to use the HTML code they give you. The only problem with that is… they don’t provide you with the ability to split test if you do that!

My method solves that problem. I’m able to use any HTML opt-in form from Aweber (or any autoresponder) and test using MuVar.

Sounds easy enough, right? Just use MuVar to test an opt-in instead of a sales page…

It’s not!
Testing an opt-in form is different!

There are other variables which aren’t used in the common sales page. I’ve managed to figure out a way to use MuVar to test my opt-in forms. And, since some of you were interested, I went ahead and wrote out detailed instructions on how you can perform these tests too!

So…

For the next 48 hours, you can gain access to this how-to guide for just $20.

After the 48 hours is up… I will not be offering this special again (at least not in the near future). This will not be a part of my list of other products.

Why wait for a split test to improve your conversion ratio when you can just test everything at once? Don’t you want more subscribers?

All you need to do is click the link below. You’ll be sent to a page with a PayPal order button. After your order is complete, you’ll be forwarded to the guide.

Click Here To Get Access To The Course

You have from now until 12 noon PST Wednesday to gain access.

- Aaron Brandon

Recently, I started running a few tests on my blog. Some initial results are in, so I want to share them with my readers.

I’ve been testing out my opt-in form. (It’s the one up there in the top right corner of this website.) Originally, I was running a simple split test using Aweber. However, as you might know, split testing is a slow and tedious process. Also, Aweber only allows you to split test while using their javascript code which makes it difficult to customize other parts of the opt-in form. There are several reasons why I don’t like split testing, but I won’t go into all of them.

Originally, my opt-in form was rather basic. It had a description, name field, e-mail field, and a submit button. But, since I wasn’t allowed to customize more specific variables in Aweber, like the submit button color for example, I had to make some changes.

So, I decided to run a multi-variable test using MuVar 2007. I use that software to test my sales pages, so why not test an opt-in form?

I decided to add some variables… here they are:

Area (the area around the form)
Headline
Description
Name
Email
Button (submit button)
Privacy (the privacy statement below the form)

I first started running a test over a month ago, but later realized I set it up wrong which skewed the data. So… I started over and here’s what has been working better so far for me (more subscribers):

Variable: Headline

I tested a few headlines for the opt-in form. One version displayed “Subscribe Now!”, another displayed “Subscribe!” and yet another displayed nothing at all. What worked best out of those three versions?

NOTHING!

That’s right, showing no headline at all for my opt-in form gave me the best conversion.

Version: 1 - Subscribe Now!
Conversion Ratio: 0%

Version: 2 - Nothing
Conversion Ratio: 3.03%

Version: 3 - Subscribe!
Conversion Ratio: 0.69%

Variable: Description

The description variable had three versions:

1. Subscribe to my blog update list to be notified via e-mail when I post a new blog entry:
2. Nothing
3. Insert your first name and e-mail to receive an e-mail each time I submit a new entry!

Here are the results:

Version: 1
Conversion Ratio: 0%

Version: 2 - Nothing
Conversion Ratio: 1.78%

Version: 3
Conversion Ratio: 2.14%

Hmm…the second version is interesting. I wouldn’t have expected someone to sign up with no description. Of course, if someone read a previous post which gave reason to enter their name and e-mail address into the form, that could have triggered it. Or, maybe someone was just curious?

Variable: Name

Two versions tested here (both were Arial, Size 1, Bold):

Version: 1 - Enter Your First Name:
Conversion Ratio: 1.33%

Version: 2 - First Name:
Conversion Ratio: 2.09%

Variable: E-mail

Version: 1 - Primary E-mail Address:
Conversion Ratio: 1.77%

Version: 2 - E-mail:
Conversion Ratio: 1.24%

Variable: Submit Button

This was interesting…

Version: 1 - Button: “Subscribe Now” Border: 5px edit
Conversion Ratio: 0%

Version: 2 - Red Border
Conversion Ratio: 2.78%

Version: 3 - Blue Border
Conversion Ratio: 0%

Version: 4 - Grey Border
Conversion Ratio: 3.33%

Version: 5 - Dark Red Border
Conversion Ratio: 0%

Version: 6 - Dark Blue Border
Conversion Ratio: 0%

Why did only a red border, and a grey border work best!? Well, we will never know the answer for sure, but I have a few guesses. My original subscribe button before doing any testing was a button with a grey border. Therefore, maybe more people signed up because they were more familiar seeing it more often? Or maybe because grey just fits in better with the theme of the website? As for red… it stands out! Some people have said that it acts like a big red stop sign. It really does stand out, so it makes sense to me.

As for the text on the submit button, I didn’t have any other versions. All of the above versions display “Subscribe Now”.

Variable: Privacy

Should you show a privacy statement? Yes? No? Apparently it doesn’t matter much (at least at this point in the test):

Version: 1 - I respect your trust and privacy.  I will NEVER share your email address with anyone at any time.
Conversion Ratio: 1.58%

Version: 2 - Nothing
Conversion Ratio: 1.69%

 - Aaron Brandon