SPLIT TESTING: My Brain Freezes Just Thinking About That Stuff...

[magentawave]

I don't do enough tracking and testing and my eye's glaze when there is the mention of 'split-testing' ...but I know I need to learn how to do that stuff.

Can anyone refer me to some SIMPLE but effective information that shows me exactly how to set all that stuff up please?

Thank you very much!
Steve

[magentawave]

Ooops, I forgot to hit the 'instant email notification' button...

[kenj]

I agree

Split testing??

What does it mean?
How do you do it?
Is it legal?
Can you get addicted to it?

So many many many many experts use the phrase "spilt testing" but so far I have never seen a clear explanation of what you do for this and who you do it to.

I will watch this post with great interest.

[Justin Jordan]

Originally Posted by kenj

I agree

Split testing??

What does it mean?
How do you do it?
Is it legal?
Can you get addicted to it?

So many many many many experts use the phrase "spilt testing" but so far I have never seen a clear explanation of what you do for this and who you do it to.

I will watch this post with great interest.

Split testing is just using two versions of something and seeing what results you get. You want to test a headline - you funnel traffic equally to two headlines and see which one gets more conversions. And so on.

Actually doing the tests can be tough, but the concept is pretty simple.

[magentawave]

Originally Posted by kenj

I agree

Split testing??

What does it mean?
How do you do it?
Is it legal?
Can you get addicted to it?

So many many many many experts use the phrase "spilt testing" but so far I have never seen a clear explanation of what you do for this and who you do it to.

I will watch this post with great interest.

I THINK one aspect of 'split-testing' is when you have a sales page that might alternate showing two different pages with very slight differences between the two of them. I think. For example, you might keep everything about the two pages exactly the same, but slightly (or totally) change the headline with the other one.

Another way (I think) is to have two identical emails with different text in the subject line.

Okay, so if I have that right about my example of split-testing, my question is how do you actually set up this stuff?

Steve

[kenj]

Hi justin

Your answer is I am sure very helpful but I do not understand it.

HOW do you do this.

I can set up a sales letter page. I can install a wordpress blog with or without fantastico. I can write articles that have a clickthrough rate of 60%. I am making an online income. BUTTTTTTTT!!!!!!!!!

Split testing - now that I do not know the HOW about that.

[magentawave]

Originally Posted by Justin Jordan

Split testing is just using two versions of something and seeing what results you get. You want to test a headline - you funnel traffic equally to two headlines and see which one gets more conversions. And so on.

Actually doing the tests can be tough, but the concept is pretty simple.

Okay cool, I think I get the concept but where can I find the mechanics for how to do this stuff? Like how would you set up your home page so that TWO people coming from organic searches will see two different sales pages? Same goes for the traffic my Clickbank affiliates send my way too? And if you have two different home/sales pages, would one of them have some no robots-no follow-no index tag so the search engines don't hit me for duplicate content? Inquiring minds want to know man! :-)

Steve

[Justin Jordan]

Well, with a webpage you can (and I've never done it) create two versions of the page and use a script installed on the page to show each visitor either one or the other. If you search the forum for 'split testing script' I'm betting you can find one that's recommended and free,

So you can have, I dunno, one page with a blue background and one with a white background and see if it makes any difference in terms of sales. Or you can different sales copy, or whatever.

Most Pay Per Click programs, like Google AdWords, have this capability built in - you can create two versions of your ad and google will alternate them and you can see which one works better.

The idea is to compare two version of whatever it is you want to test to see which gets better results. You can do it with virtually any aspect of a webpage or whatnot, so long as you get some traffic.

There's a very good ebook called, if I remember right, 'Small Changes, Big Profits' that will tell you everything you ever wanted to know about split testing.

[Justin Jordan]

Originally Posted by magentawave

Okay cool, I think I get the concept but where can I find the mechanics for how to do this stuff? Like how would you set up your home page so that TWO people coming from organic searches will see two different sales pages? Same goes for the traffic my Clickbank affiliates send my way too? And if you have two different home/sales pages, would one of them have some no robots-no follow-no index tag so the search engines don't hit me for duplicate content? Inquiring minds want to know man! :-)

Steve

Too technical for me - I can tell you what it is but I can't even begin to tell how to do the tech stuff you'd need. Sorry - me are am caveman.

But here are some threads with helpful stuff:

Split Testing...how to?

Simple Split-Testing Script from Ads4Dough


That first thread actually talks about Paul Hancox, who wrote 'Small Changes, Big Profits' and will probably help a lot.

[John Ritz]

Here's a guide that gets into both split testing and multivariate testing, and much more:

http://www.copywriters-toolkit.com/profitdna.pdf

(Right-click link, then "Save As...")

Feel free to pass it around or send it to your list.

Enjoy!

John

[kenj]

John
I really want to thankyou for your ebook response.

This is what makes this warrior forum the best in the world

Unbelievable

[MarcusXavier]

the best way to split-test is to invest in some type of tracking software like Adtrackz or my personal weapon of choice, Conversion Prophet.

If you can't afford either, just make a few affiliate sales and re-invest.

IM is a business and you must treat it as one. So this means that you need to re-invest in your business. A one-time investment in tracking software will yield and save you thousands. Literally.

[chris_surfrider]

Google Optimizer is free and it can run split tests, multivariate, etc.

I use a very basic A/B split tester but I'll probably switch to Google soon.

Split testing is EASY.

Just choose ONE thing to test. Such as a headline - or an entirely different salesletter.

Try to get at least 50 actions (sales), and then tally up the score and go with the winner.

Then keep on split-testing elements of the winner (your "control") by testing different headlines, different layouts and so on.

Again, try to eliminate the "randomness factor" by letting each test run until you've got about 50 sales/actions combined between both versions of the page.

Wash, rinse and repeat, and watch your conversions climb.

You'll kick yourself for not doing this earlier.

-Chris

[jhongren]

Hi Chris,

I am doing split-testing now but I am not sure if I am doing the right way. I am lucky to have friends with big lists, above 2000 per list. Then, they drive them to my sales page using a promo email I sent to my friends.

In your own opinion, is this method viable too?

Beside headlines, what else can we test? (bonuses, sub-headlines etc... ?)

Would love to hear from your experience.

P.S: By the way, I love your past threads and posts. They are really good.

Cheers,
John

[Daniel E Taylor]

I like to test 2 different sales copies completely.

Then I test video vs video.

Google website optimizer has all the tools needed for free
to run your split testing.

And with Google Analytics Linking with Your adwords campaign
and Google Conversion tracker I don't see why you would need
anything else.

And they have the multivariate testing as well with like
up to like alot of different small changes.

[David Raybould]

Hi guys,

Chris S is right - the Google website optimizer is super easy to use. I was using some third party all singing all dancing software which has a monthly fee, but I recently switched to Google's tool, because it's free, and it does a much better job.

Check it out under business solutions from the main google home page. It's easy to install, just create some variations of the same page (different headlines, different offer etc) and then paste the code that Google gives you into your pages.

They'll then rotate them and keep track of actions, and calculate the whole thing into percentages, so you can clearly see which is the winner.

Split testing is VERY powerful. I just started testing a new squeeze page I've been working on, and with the very first test I was able to boost the conversion rate from 18% to about 33%. That's almost DOUBLE the subscribers for just a small change, and I've still got plenty of stuff left to test.

Another example - I did a broadcast to a list of mine yesterday, and I split the list into two, and sent the same email to each with a slight change in one of them - I added seven extra words in the body copy that told the reader exactly what action to take when they reached the page I was referring them to. The difference? The email with the extra instruction had a click through rate over 25% higher than the other one...and that makes a big difference to the bottom line...

Do you see how powerful testing can be?

Jhongren - you can test all kinds of things, pretty much everything on your page is going to be influencing the sales decision of your buyer. Some examples are: Headline, subheadline, lead into sales letter, font, font size, font color, background color, offer, guarantee, and price, to mention but a few. I feel pretty confident in saying that if you systematically tested all of those things against some variations you'd boost your conversions significantly.

Hope that's useful.

David

[jhongren]

Jhongren - you can test all kinds of things, pretty much everything on your page is going to be influencing the sales decision of your buyer. Some examples are: Headline, subheadline, lead into sales letter, font, font size, font color, background color, offer, guarantee, and price, to mention but a few. I feel pretty confident in saying that if you systematically tested all of those things against some variations you'd boost your conversions significantly.

Wow, David...thanks for your advice. Is there a point we stop testing? I know testing is good...but how do we know when do we stop?

Thanks in advance,
John

[MikeHumphreys]

John Ritz's Profit DNA is a rock-solid free report that I frequently recommend reading. Another paid resource is Paul Hancox's Small Changes Big Profits ebook Small Changes: Big Profits (2008 Edition) (non-aff link)

Google's tool is popular because it's free and works well. The catch is you are sharing all of your business data about conversion rates, sales, and best performing keywords with Google.

Personally, I don't like anyone else having access to my testing data so I had my own multi-variate testing tool built in 2007 called Easy Multi Tracking. It's been regularly updated with a lot of great features which you can check out here: Easy Multi Tracking: Professional Copywriter Reveals His Proven Conversion Boosting Tool That You Can Use To Skyrocket Your Conversion Rates FAST!

The sales letter for Easy Multi Tracking has a video on it that shows you how to set up a multi-variate testing campaign.

Hope that helps,

Mike

[Easy Cash]

Yes - Recommend Googles Optimizer.
This is easy to set up - just a bit of code and away you go.

Have you tried using this?

It really is easy.

[MikeHumphreys]

Originally Posted by chris_surfrider

Just choose ONE thing to test. Such as a headline - or an entirely different salesletter.

Try to get at least 50 actions (sales), and then tally up the score and go with the winner.

Then keep on split-testing elements of the winner (your "control") by testing different headlines, different layouts and so on.

Again, try to eliminate the "randomness factor" by letting each test run until you've got about 50 sales/actions combined between both versions of the page.

Wash, rinse and repeat, and watch your conversions climb.

You'll kick yourself for not doing this earlier.

-Chris

Chris, good post but 50 actions isn't always enough testing data. 100 Actions is the minimum number that I recommend using. The reason why is you want to make sure your testing doesn't flip-flop. In other words, it needs to have a high level of statisical confidence/statisical significance... I recommend a statisical confidence level of at least 95%.

If your testing tool doesn't that tracking feature, then you can try this free tool: :::SplitTester.com:::


Hope that helps,

Mike

[magentawave]

Thank you everyone for your input regarding split-testing here...and the link to the free ebook too!

Does the Google Optimizer give you the code so that two different organic searchers will see two different sales pages? Or does the Google Optimizer only work with your Adword campaigns?

Thanks again.
Steve

[chris_surfrider]

Hi Mike, re: 50 actions

While I agree with you that 100 actions is certainly the better route to take as opposed to 50, that may be out of reach for someone who's just starting with this and has limited traffic or seed capital.

In terms of scientific "significance" (where one can have confidence that a pattern has emerged that is the result of the copy, and not just proverbial coin-flipping), determining significance simply depends on your acceptable risk level.

As with any test, sample size is the crucial factor in determining confidence, but it's not the ONLY factor.

Others are the price point, the potential for reach, and so on.

If you're selling a $1000 course, obtaining 100 sales for testing purposes could be difficult.

However, it's a different story for a $7 report.

Anyway - for anyone who's interested in what the scientific community uses as a formula for determining significance, check out Statistical significance - Wikipedia, the free encyclopedia

And Statistical hypothesis testing - Wikipedia, the free encyclopedia

As you'll soon see, the rabbit hole goes down pretty deep, and that there's a lot of argument and criticism about the various formulas for determining significance.

The way I see it, you just gotta find what works best while being realistic about your sample size, all things considered.

If you're basing your "conclusion" on 150 visitors, that's not testing.

However, after 10,000 visitors - you should definitely begin to see a trend.

Thanks,

-Chris

[Kenneth L]

Split testing is where the money is made!

You need to do it!

Google WebOptimizer is a good place to start!

However Conversion Prophet is definitely the best script if you want to improve conversions FAST!

[tommyp]

Well there's regular split testing and there's multivariate split testing and some other ways. The simplest is A/B testing. You can do it with Google now so I would look into it.

You may be confused because it's a foreign concept and you read a bunch of different things about it but it really is very simple to understand the basics of it.

You can use Google or you can get a simple split testing script (maybe for free).

When you split test between two pages or ads or whatever to see which works better try to make only one change otherwise you won't be sure what made the difference. For example if using it on a sales page and you are using two versions to test against each other you may want to start by testing to see what headline converts better. So the only difference on each page should be the headline.

[Paul Myers]

I haven't looked at Mike's multivariate script, but I hear very good things about it.

I do have reservations about giving Google - or anyone else - that kind of information about any important site.

The mechanics of split testing are going to vary based on which product you use and what type of tests you run. For those details, Read The Fine Manual. For really simple split testing, Gary Ambrose' "Tracking Ace" is more than sufficent for a beginner.

For more advanced testing, you'll need something ... well ... more advanced.

My advice is to get anything you feel you can understand and start doing any sort of test at all. When you see the differences that variables can make, sometimes even very small changes, you'll get addicted. And you can learn it a little bit at a time.

It's worth the effort to learn. Just do it one step at a time.


Paul

[onlinemoney00]

It depends on what you are split-testing. In internet marketing terms, all split testing is putting two variables (e.g Headline) together and finding out what one performs best. You can use google analytics tool and g oogle conversion tool, these are totally free especially when you are using adwords to get visitors to your site. Google also explain in detail how to use these tools

[John Ritz]

I haven't looked at Mike's multivariate script, but I hear very good things about it.

I bought it from the beginning, and I highly recommend it.

I use it, and for what Mike charges for the software, it's steal compared to the value.

But that's just me. I'm a testing freak sometimes. On teleseminars and such, I frequently talk about some tests I've done and their results. Results you can plug in and take to the bank.

A lot of folks have asked me about publishing my results in a guide. The problem is, one it hits printed form, it'll be circling the Internet for a L-O-N-G time to come.

And test results are a fickle thing. What works today might not necessarily work tomorrow (or at least as well), so I try to avoid putting my results in stone.

However, I'm happy to talk about it whenever asked.

And Mike's tool is a big part of these successes!

Cheers,

John

[Dan Grossman]

Google Website Optimizer lets you easily set up split tests and many more complex types of experiments.

It's free of course.