Split testing lessons learned ?

KenCalhoun · Very Active Member Joined: 08 Dec 2004 Posts: 154

I just had a basic split test script coded and am going to start split testing all my salesletters at all my "money sites".

Before I go down that path, any lessons learned, the "most important stuff to get right" before I start split testing headlines, closes, bonus stacks, price points and the rest?

Appreciate any feedback from folks who've split tested heavily over the last 2+ years..

For example, in one of my soon to be launched sites' tours, I have a flash animation:

http://www.swingtradinguniversity.com/swingtrading_tour01.htm

as eye candy (stock traders like flashy lights, my flash format short copy site did $302K in sales one year)... and will split test that landing page against a long copy benefit rich page..

Let me know, any tips on split testing different versions of sites, example multiple versions of name squeeze to long copy sequence, vs long copy only, vs flash/corporate hybrid to long copy for specific products, lots of variations to play with. (btw I just created that entire site "start" from scratch starting today at 10am)..

thanks,

ken
_________________
http://www.WritingAdwords.com <-- Profit-boosting tactics for Adwords advertisers!

Roosevelt Cooper · Posted: Thu Dec 09, 2004 6:10 am Post subject: Split Testing

Ken,

I haven't been split testing for 2+ years (haven't even been in internet marketing for a full year yet). However, I can tell you that spilt testing two completely different web sites isn't really going to tell you much. John Reese talks about this in his course and the key to split testing to get valid data that you can make business decisions on is to split test small changes (i.e. Two web pages with everything the same except the top headline).

The problem with split testing two completely different web sites is while you know which one is the winner, you don't know why. For example, let's say you run your test and the flash site converts 2% and the long copy converts 1%. You are going to say the flash site is better and use it.

However, what you don't know is that the long copy site didn't do as well simply because the headline wasn't that great. You also don't know that if you would have just change the headline, you would have gotten a 4% conversion rate.

This is why you must test small changes in order for split testing to be effective. Testing small changes lets you know the reason WHY a particular set of copy was the winner.

Hope that helps!
_________________
Roosevelt Cooper

Feel free to click on my profile name to send me a message if you have any questions or need my assistance!

drumdance · New Joined: 13 Dec 2004 Posts: 6

Ken,

Roosevelt is dead-on. Here's an example of a split test I did:

Page A: http://microurl.com/cts/mandarin1
Page B: http://microurl.com/cts/mandarin2

Notice that the only difference between the two is the Flash audio applet.

Good luck!

-Derek
_________________
Derek Scruggs
Click Thru Stats - http://www.ClickThruStats.com
Fan Prints - http://www.FanPrints.com
Happy Mudan - http://www.HappyMudan.com
Ask me a question - http://www.AskDerekScruggs.com

FrankKern · New Joined: 12 Dec 2004 Posts: 6

Which one won?

The only advice I can offer is to take pencil to paper and CHECK the results yourself.

I've found inaccuracies in every split testing program I've used. That's why I endorse the "Cave Man Method" where you write down the clicks for A, write down the sales for A ...and do the math.

Then do it again for B.

It's a hassle but that's some important data.

NickTheGeek · Posted: Mon Dec 13, 2004 1:23 pm Post subject:

I've mentioned this trick on the TS forum but I'll restate it here:

To minimize any "skewing" of data when split testing (especially price) name the tested pages the same but give them different extensions, like:

http://www.yourdomain.com/special.htm
http://www.yourdomain.com/special.html
http://www.yourdomain.com/special.shtm
http://www.yourdomain.com/special.shtml

By naming split test files in this way, rather than "index1.html", "index2.html", etc. it stops people looking to see if there they can find other pages where they might be able to get your product or service cheaper ...

This will definitely work with split testing 2 pages at a time if you use ".htm" and".html" extensions

You *might* be able to use the other two extensions as well, depending on how your server is setup ... the best way is just to try it ...

You can make up your own extensions but you'll need to modify your webserver's ".htaccess" file which should only be attempted by proper "propellorhead", "geek" or someone who really knows what they're doing ... Wink

Cheers

Nick Smile

_________________
Download my Free Special Report PDF Showing How to Create High-Priced Video Products - Just Like the Pros - Absolutely FREE! Comes with Resale and Redistribution Rights. http://www.streamingvideosecrets.com

Derrick Pizur · Posted: Mon Dec 13, 2004 3:58 pm Post subject:

Great advice Nick, I didn't think of that.

Derrick

drumdance · New Joined: 13 Dec 2004 Posts: 6

Frank,

The audio page won, but not by much. It took quite a few clicks to get to a meaningful conclusion. And even then I'm not positive. I want to do another test with removing the first link to eBay. Some people may be clicking that before they read/listen to the rest of the page.

The little things matter. Very Happy

_________________
Derek Scruggs
Click Thru Stats - http://www.ClickThruStats.com
Fan Prints - http://www.FanPrints.com
Happy Mudan - http://www.HappyMudan.com
Ask me a question - http://www.AskDerekScruggs.com

NickTheGeek · Posted: Tue Dec 14, 2004 3:33 am Post subject:

Mal Keenan · Contributor Joined: 13 Dec 2004 Posts: 13

Could someone please recommend a simple easy to install script that I can use for split testing? Im thinking mainly of testing my optin form.

I've tried clickalyser and it was too complicated for my basic needs.
_________________
Mal Keenan
http://www.home-business-tips-newsletter.com/

A Simple, Step-by-Step Plan For Earning
$7000 Per Month In Dependable, Residual Income. Don't miss: http://www.eliteteamireland.com

NickTheGeek · Posted: Tue Dec 14, 2004 4:01 am Post subject:

A really good free one is the Scientific Internet Marketing Assistant (SIMA)

You can normally get it from here:

http://www.onlinemarketingtoday.com/software/split-testing/

The link for it has been removed as he's having the software recoded, but I've uploaded a copy of the last version to my server here:

http://www.ntgserver2.com/downloads/sima.zip

It's written in PHP, is really easy to install and doesn't require any MySQL databases to be set up ... only to change the permissions (CHMOD) on a couple of folders to 777 ... full installation instructions are included ...

I hope that's helpful ...

Cheers

Nick Smile

_________________
Download my Free Special Report PDF Showing How to Create High-Priced Video Products - Just Like the Pros - Absolutely FREE! Comes with Resale and Redistribution Rights. http://www.streamingvideosecrets.com

Mal Keenan · Contributor Joined: 13 Dec 2004 Posts: 13

Thanks a lot Nick

You're a gentleman Smile

_________________
Mal Keenan
http://www.home-business-tips-newsletter.com/

A Simple, Step-by-Step Plan For Earning
$7000 Per Month In Dependable, Residual Income. Don't miss: http://www.eliteteamireland.com

NickTheGeek · Posted: Tue Dec 14, 2004 9:28 am Post subject:

Alan Saltz · New Joined: 14 Dec 2004 Posts: 7 Location: New York

Okay... I'd like to play devil's advocate for a little.

I know you're supposed to split test one thing at a time... but who is to say that a totally different style (short copy, selling by educating, etc.) won't also significantly improve conversion.

What's the harm in testing 2 totally different sites if the new one significantly improves response.

THEN you can start tweaking the headline of that new page, adding audio, etc.

While it's always said that a new headline can boost response by 19 times, or whatever the statistic is... some copy just needs a total overhaul.

I'm really curious to hear what people think about this (especially because I'm guilty of what I speak of!)

Any thoughts greatly appreciated.

Best,
Alan Saltz
_________________
Simple Yellow Pages Advertising Techniques That Boost Response...
http://www.YellowPagesProfit.com

Mal Keenan · Contributor Joined: 13 Dec 2004 Posts: 13

Hi Alan
I dont see anything wrong with this at all and is what I will be doing when I get my split testing going.

After all the whole point of split testing is improving response and sales. So if a sudden insight caused me to make big changes that proved fruitful then great. I could then work on the better copy to improve it even more with small changes as suggested by others.

Just my opinion of course, I havent even started yet. Smile

_________________
Mal Keenan
http://www.home-business-tips-newsletter.com/

A Simple, Step-by-Step Plan For Earning
$7000 Per Month In Dependable, Residual Income. Don't miss: http://www.eliteteamireland.com

Derrick Pizur · Posted: Tue Dec 14, 2004 6:39 pm Post subject:

I think small changes is still best. Unless you are 100% sure a major change(s) will improve that much.

If you do make major changes how will you know which change caused the increase in sales? <- Im asking here(looking forward to your response).

Derrick Pizur