|
View previous topic
::
View next topic
|
| Author |
Message |
oran
Joined: 11 Aug 2003 Posts: 9 Location: Fort Bragg CA USA
|
Posted: Fri Oct 03, 2003 12:31 am Post subject: Split testing |
|
|
Does anyone use split testing on their SBI site(s)? If so, how?
Thanks,
Oran _________________ Find a great greenhouse the easy way: www.great-greenhouses.com |
|
| Back to top |
|
 |
Debs
Joined: 16 Aug 2003 Posts: 4296 Location: NY
|
Posted: Fri Oct 03, 2003 1:38 am Post subject: |
|
|
Maybe explain what split testing is? I have never heard the term before and I have been around a bit. **clicking her dentures and combing what's left of her gray hair**
Debs _________________ Learn how to turn keyphrases into quality, well-targeted articles your visitors and SE's will love with Gary Antosh's new ebook "Web Content Made Easy!" |
|
| Back to top |
|
 |
Robert
Joined: 27 Sep 2003 Posts: 57 Location: Southern California
|
Posted: Fri Oct 03, 2003 2:14 am Post subject: |
|
|
Debs - I may be wrong, but I think split-testing is when you have two versions of a web page and you send half of your visitors to each page and then see which page provided the best results.
Oran - As far as how it's done, I do not know. _________________ Bob Hampton
http://www.bankruptcy-basics.com
Download our free "Bankruptcy Basics" eBook. |
|
| Back to top |
|
 |
Debs
Joined: 16 Aug 2003 Posts: 4296 Location: NY
|
Posted: Fri Oct 03, 2003 3:33 am Post subject: |
|
|
Ok, now I understand thank you Robert. It sounds like it would require a redirect of some sort, which could possibly be in the html of a page itself ... maybe javascript so it would alternate between page 1 and 2 ... this could be done with SBI just by uploading the redirect page with the javascript in it or putting the javascript in the building block if no requirement exists to have it in the <head> section of the page.
Then if you promote the redirect page via PPC ... anyone who clicks the PPC link would go thru the redirect page and visitors would alternate between pg 1 and pg 2.
The redirect page is not one you would want submitted to regular search engines however ... as they wouldn't index it, and therefore wouldn't send any visitors your way ... they could, however, end up indexing either pg1 or pg2 if the redirect shows any link on its page.
The other concern I think would be having pg1 or pg2 indexed ... during the period you experiment, you would likely want to have pg1 and pg2 as noindex so as not to skew your results.
Once you determine the page that has the best results, then you could remove the noindex tag, and link to your page normally, as well as submit to SE's openly.
Ok ... now you want to know where to get the javascript to do this? I have no clue however some great resources would be:
http://javascript.internet.com/
http://www.javascriptkit.com/cutpastejava.shtml
http://www.dynamicdrive.com/
You could also check rentacoder and see what it would cost to build a custom script if you don't find what you need already made.
Now I would like to make a couple points here, so bear with me
SBI does not offer access to the server, or support our use of CGI, so some of the more mainstream ways to split test are not available to SBI users.
This is why I recommended javascript as a possible solution as it goes into the html page itself. I also suggest that if you use this option you set all 3 pages (the redirect page you are going to advertise, pg1 and pg2) to noindex. The redirect page would likely not get indexed, but SE's might try to index pg1 and pg2. Indexing of pg1 or pg2 could have detrimental effect on your testing because visitors would be able to go to them directly, and you would lose control of the split.
The other concern with pg1 and pg2 being indexed is if they are substantially the same, some SE's may pick them up as duplicate content, which could harm the ranking of your entire site. Noone wants to risk that I don't think.
So that kind of leaves PPC as the suggested method for testing. I have no idea if the PPC's would permit this kind of setup for split testing since I don't do any PPC advertising.
Another alternative would be to consider using a tracking link from the tracking library, set your PPC ad, let it run for 7 days, then switch your content on the page for the next 7 days while the same ad is running. This, again, with noindex on the page.
Hope this helps if not ... I tried darnit! LOL
Debs _________________ Learn how to turn keyphrases into quality, well-targeted articles your visitors and SE's will love with Gary Antosh's new ebook "Web Content Made Easy!" |
|
| Back to top |
|
 |
Charlie
Joined: 22 Aug 2003 Posts: 3305 Location: UK
|
Posted: Fri Oct 03, 2003 12:15 pm Post subject: |
|
|
| Robert wrote: | | I may be wrong, but I think split-testing is when you have two versions of a web page and you send half of your visitors to each page and then see which page provided the best results. |
That's just how I'd describe it!
When it comes to implementation, the traditional way is to use a CGI (or PHP) script to do the page display and tracking for you. (There are many of these available for free or very low cost.)
The way I see it, there are (at least!) 3 problems...
1. SBI! doesn't allow server side scripting. One solution would be to try and have a second domain (.net?) - with server side scripts - for pages like these. This would be messy and complicated, though.
2. Although not quite as bad as with the client-side Javascript solution Debs gave, the SEs aren't crazy about it. This is important when a page gets a lot of traffic (the cases where split testing are most worthwhile). From a link popularity point of view, the problems are grim.
3. As with all tracking, it's nice to go as far as the sale, but without help from the merchant, you will lose touch of visitors once they click an affiliate link. Unfortunately, one version of a page may send more visitors to a merchant, but less of them may buy than from the other version of the page. Without the merchant's help, you risk jumping to the wrong conclusion.
Everything considered, the best compromise might be to run your split tests using the PPCs and hosting your pages on a test domain (with no SE indexing). You could always link to the main SBI! site from here. Only when you'd got conclusive evidence from a sample of at least 1000 visitors should you consider changing the real page's control copy.
Although the visitors would be a slightly different type than those from the free SEs, this might be the best compromise. Potential problems with tracking as far as the sale, would still be the same.
Hope this helps,
Charlie. _________________ "Before I speak, I have something important to say."
- Groucho Marx |
|
| Back to top |
|
 |
SgtAw
Joined: 02 Oct 2003 Posts: 64
|
Posted: Sat Oct 04, 2003 2:59 pm Post subject: Script for split testing... |
|
|
Hi All,
here is a split testing script from thje guy that provide SBI with the tracking tool.
http://www.cgitoolbox.com/free.html
Essentially, you put this script on another site and have it redirect to the two SBI (or whatever) pages you want to test.
When someone click on the link this script will send them to one of the two sites.
Hope this helps,
Ed |
|
| Back to top |
|
 |
Dean
Joined: 03 Oct 2003 Posts: 6
|
Posted: Sat Oct 04, 2003 3:49 pm Post subject: |
|
|
Duncan Carver offers an amazing piece of software for free that will do all kinds of split run testing. It's called the "Scientific Internet Marketing Assistant"
As SBI doesn't allow scripts you would have to set it up on another domain and redirect to it.
Grab this while he's still giving it away free.
http://www.marketing-strategy.org/
Dean _________________ Need more Laughter in your Life?
Check out the #1 Joke Book on the net....
That's Comedy! Joke Book!
http://www.ThatsComedy.com/tc-jokebook.htm |
|
| Back to top |
|
 |
SgtAw
Joined: 02 Oct 2003 Posts: 64
|
Posted: Sat Oct 04, 2003 5:53 pm Post subject: Download doesn't work |
|
|
I went to the download page and it doesn't work
Thanks,
Ed |
|
| Back to top |
|
 |
Dean
Joined: 03 Oct 2003 Posts: 6
|
Posted: Sat Oct 04, 2003 6:04 pm Post subject: |
|
|
You're right. Looks like there's a problem with his autoresponder.
I'm a subscriber to his mailing list so I'll see if I can get a message to him.
Dean _________________ Need more Laughter in your Life?
Check out the #1 Joke Book on the net....
That's Comedy! Joke Book!
http://www.ThatsComedy.com/tc-jokebook.htm |
|
| Back to top |
|
 |
Dean
Joined: 03 Oct 2003 Posts: 6
|
Posted: Sun Oct 05, 2003 2:23 am Post subject: |
|
|
It's working now. _________________ Need more Laughter in your Life?
Check out the #1 Joke Book on the net....
That's Comedy! Joke Book!
http://www.ThatsComedy.com/tc-jokebook.htm |
|
| Back to top |
|
 |
oran
Joined: 11 Aug 2003 Posts: 9 Location: Fort Bragg CA USA
|
Posted: Mon Oct 06, 2003 12:11 am Post subject: Thanks |
|
|
Thanks everybody. Thanks especially to Debs and Charlie for such detailed replies.
Dean, I've been following SIMA for a month now on their forum, but have hesitated to try it because of its problems and Duncan seems to be distracted by his new MLM. What is your experience with SIMA?
Let me redirect this thread (since I started it ). Are any of you USING split testing?
If so, how?
If not, why not?
Best wishes to all,
Oran Kangas _________________ Find a great greenhouse the easy way: www.great-greenhouses.com |
|
| Back to top |
|
 |
Debs
Joined: 16 Aug 2003 Posts: 4296 Location: NY
|
Posted: Mon Oct 06, 2003 1:39 am Post subject: |
|
|
| Quote: | Let me redirect this thread (since I started it ). Are any of you USING split testing?
If so, how?
If not, why not? |
The initial reason I didn't was money for the PPC testing. Now it is something I am going to consider doing. Not sure though how it would work since I don't sell my own products but promote affiliate programs that tie into my themed content website. Any thoughts?
Debs _________________ Learn how to turn keyphrases into quality, well-targeted articles your visitors and SE's will love with Gary Antosh's new ebook "Web Content Made Easy!" |
|
| Back to top |
|
 |
oran
Joined: 11 Aug 2003 Posts: 9 Location: Fort Bragg CA USA
|
Posted: Mon Oct 06, 2003 2:39 am Post subject: |
|
|
| Quote: | | Not sure though how it would work since I don't sell my own products but promote affiliate programs that tie into my themed content website. Any thoughts? |
Hi Debs,
The beauty of testing is you can use it for anything.
For example, what might work better for you:
a simpler 1 column home page?
different colors?
font change?
headline?
new quote?
no quote?
larger font?
Then there is the issue of how you define "work better". More pages read; newsletter sign up; affiliate link out; direct purchase; etc.
It's all up to you. What do you want prospects to do? What might influence them to take that action? That's what testing is all about.
That testing is important in undeniable. The devil is in the details.
Best wishes,
Oran Kangas _________________ Find a great greenhouse the easy way: www.great-greenhouses.com |
|
| Back to top |
|
 |
Charlie
Joined: 22 Aug 2003 Posts: 3305 Location: UK
|
Posted: Mon Oct 06, 2003 2:52 pm Post subject: |
|
|
Hello everyone.
Split testing is used most in direct marketing.
Bearing in mind that you need at least 1000 visitors for each test, to draw any useful conclusions, it is not practical for most people to test anything other than major changes - for example major design changes or different headlines.
| Debs wrote: | The initial reason I didn't was money for the PPC testing. Now it is something I am going to consider doing. Not sure though how it would work since I don't sell my own products but promote affiliate programs that tie into my themed content website. Any thoughts?
Debs |
Unless the merchant lets you track all the way to the sale (which most don't), you'll only be able to track click through rates. This may not be in proportion to sales.
The only workaround, is to join the affiliate program each time for every page you want to test. Painful!
Hope this helps,
Charlie. _________________ "Before I speak, I have something important to say."
- Groucho Marx |
|
| Back to top |
|
 |
| View previous topic :: View next topic |
|