Hi All,

I’ve recently come across two new split testing services that are trying to bring split testing to the masses.
http://visualwebsiteoptimizer.com/
http://unbounce.com/

While I love GWO, the underlying assumption is that you can create the HTML for your tests on your own.
For many of the smaller mom-and-pop sites, simply creating the test pages or sections is beyond their capabilities.
Both of these new products come with a built in WYSIWYG editor, which truly lowers the barrier of entry for split testing.

Both products look promising and you should try them out.
One thing I really liked about Visual Web Optimizer is that it will grab your current page and then let you create variations based on your existing page.

Both products are in private beta but I got a hold of some invite codes.

The invite code you can use for Visual Website Optimizer is carsonified

For Unbounce there is a special sign up page for prusak.com readers: http://unbouncepages.com/private-beta-prusak/
Don’t forget to use PRUSAK as the sneak peek code when signing up for Unbounce.

Enjoy and let me know what you think.

- Ophir

I was reading a simple case study today.
They were testing two different versions of a banner that was advertising a webinar.
One of the banners had an image of the presenter, while the other did not.
The banner without the image of the presenter won (by over 50%).

One of the comments was something along the lines of:
I guess this audience prefers banners without an image of a person.

*sigh*

If you don’t immediately realize the mistake the commenter made, don’t feel bad. It’s a very common mistake.

Beyond the fact that a specific banner (which did have have an image of the presenter) won over a different specific banner (which did not have an image of the presenter) you really can’t be sure of anything.

The loosing banner might have won with:

• An image of a different person
• A different image of the same person
• The same image of the same person in a different position or size on the banner.
• The same image of the same person in the same position and size but with different elements on the banner changed.

The point is:

Don’t jump to generalized conclusions based on the outcome of a specific experiment.

-Ophir

I just found this post from a couple of months ago that has a plethora of great examples of call to action buttons.

Changing the call to action button is one of easiest ways to improve your conversion rate.

Check it out:

http://www.smashingmagazine.com/2009/10/13/call-to-action-buttons-examples-and-best-practices/

Google Analytics just officially announced a boatload of new features.
Instead of writing about them myself, I’ll let you check it out straight from the official blog:
http://analytics.blogspot.com/2009/10/google-analytics-now-more-powerful.html

I’ve known about the new features for a while (under NDA) and have been thinking about how to best take advantage of them.

More to follow in future posts.
- Ophir

Jeremy Aube from ROI Revolution just posted an awesome solution for not needing to update your conversion page tags on your GWO conversion pages.

Check it out:
http://www.roirevolution.com/blog/2009/10/universal_conversion_code_for_google_website_optim.html

Out of the box, MVT experiments with GWO will serve up all possible permutations for an experiment.

Sometimes you want to create sections that are related to each other. For example - a header and footer section that need to be coordinated with each other. The red header needs to be displayed with the red footer and the blue header needs to be displayed with the blue footer.

There is a very simple solution to this.

Just create the experiment as usual and right after you launch the experiment use GWO’s pruning function to remove the combinations you don’t want to serve up (i.e. the red header with the blue footer and the blue header with the red footer).

Enjoy!

Every now and then I find a good case study.
This posting serves as a simple list of those I want to keep for further reference.
I’ll be adding posts as I find them.

Enjoy

http://dustincurtis.com/you_should_follow_me_on_twitter.html
http://www.37signals.com/svn/posts/1525-writing-decisions-headline-tests-on-the-highrise-signup-page
http://carsonified.com/blog/business/how-to-increase-sign-ups-by-200-percent/
http://whichtestwon.com/
http://www.edrivis.com/?p=211
http://www.seointern.com/blog/jack-russell-terrier-raises-conversions.html
http://www.conversiondoctor.com/conversion-blog/2007/usability-split-test-results-link-appearance-matters-more-than-you-think/
http://thinkeyetracking.com/Blog/?p=165
http://www.blastam.com/broadcast/2009/06/google-website-optimizer-increases.html
http://www.the-search-engine-optimizer.com/newsarchive/698ConversionIncrease.pdf
http://www.slideshare.net/a4uexpo/how-to-increase-your-conversion-rate

I received an email last night to my gmail account (my browser is chrome).
Here is what it looked like:

I wanted to learn more, so I tried clicking on a few things, but nothing was clickable.

I then turned on images to see what I was missing and got this version:

See the difference?

The ONLY call to action in the email is the learn more button at the bottom which is an image!

While it might make sense having the call to action being an image if the entire email is an image (which is not always a good idea) having most of the email text and ONLY the call to action be an image is very confusing.

Also, you should always preview your email blasts with images off before sending.

Just a quick update.

For the past few days, the WP-Supercache plugin was causing site issues.
This has now been resolved.

- Ophir

One of the limitations of using a standard A/B split test is that it won’t work out of the box if the page you are testing is the “action” page for a form that uses method=POST.

For example, lets say you have a simple form page - form.html that looks something like this:

Address <input type=text name='address'>

You’ve created process_form_2.php to split test against process_form_1.php.

The problem is that when process_form_1.php redirects to process_form_2.php, you will have lost all of your form data since your form uses POST.

There are two solutions for this:

1. Switch from method=POST to method=GET.
If this is an option, it’s probably the easiest solution.

2. Setup the experiment as a multivariate experiment and change the form action line in the HTML.
For example:

<script>utm_section('form action')</script>

</noscript>

and then in the experiment setup, add a variation like this:

<form action='process_form_2.php' method='POST'>

Hope this helps.
Ophir