The Importance of Split Testing a Prospect Conversion System

Spending ad money online? Sometimes small changes can make all the difference in the World to how people react to your content. The importance of testing a prospect conversion system should not be overlooked as this should significantly increase your bottom line and can reduce advertising expenses.

Split Testing (A/B Testing) basically takes a variation of a landing page and compares it to another.  Multivariate testing focuses on individual elements of a page. You can use the A/B Theme Testing Plugin for WordPress to actually change the elements of a page on a specific themeset duplicate and see how changed elements react. Entirely different themes can also be tested. A nice easy way to have tests organized and track changes can be by theme vs. going page by page.  There’s a nice example of how this plugin has been used at WPMU DEV to do A/B Split Testing with Google Analytics on the WPMU DEV site.  It’s also important to take into account the differences in sidebars when split testing.

Split Test Prospect Conversion
"No. No. It's a small men's carryall."

Facts About Split Testing

  • Separates Seasonal and External Factors
  • Simultaneous Data is Considered Most Accurate
  • Results Can be Accelerated
  • Trying New Ideas is Pertinent for Success

Why A/B Test

  • Different Audiences React Differently
  • Small Changes Can Be Significant
  • If People React Positively, Your Site is Easier to Use
  • Increase Revenues from Main Conversion Points
  • Lessen Bounce Rates

Some Examples of Actual Tests

Highrise Split Tests by 37 Signals on Highrise
Human Images Affect Conversion

See Group of Tests Done by 37 Signals

Button Color Split Test on the Daily SEO Blog
Which Color Draws You In?
Split Test Screencapture
Example of a More Extensive Theme Test

Dmix Case StudyAquasoft Sales Page on SEOmoz Blog, CRO Post, 21 Tests Done

Changes Make a Difference

Add to Cart Split Test
Capture Shows Potential Button Test

An Example of How to Change Add to Cart Button Size for Testing

Examples of What to Split Test For

  • Same Theme, Different Elements and Structure
  • Completely Different Themes
  • Seasonal Graphics and Related Text
  • Price Changes, Discount Effectiveness
  • Prices Shown With or Without Tax
  • Graphics Impact to Conversion
  • Text Block Positioning
  • Banner or Non-Banner Usage
  • Font Styles and Colors
  • Banner Vs. No Banner
  • Personal Photo Usage Vs. No Photo Usage
  • Different Buy Now or Add to Cart Buttons
  • Newsletter Subscription Box Formats
  • Ad Text or Graphic Impact on Conversions
  • Large Content Structure Vs. Simple
  • Number of Blog Posts, Excerpt Length

This is a great article to help recognize deficiencies that effect conversion rate.  These issues can be cured on one side of the testing and left original in the other to determine if this is making a difference on your site.

Those are just some ideas, depending on the site setup and product/service offering, just about every element can make a difference.  Mountains of data from other site owner a/b test results can be found on the internet for ideas related to changes that have increased conversion.

Google Website Optimizer and WordPress

Initially, many were using Google Website Optimizer for Split Testing WordPress, but the setup for this became somewhat tedious. It is still a viable solution but somewhat more tedious without a plugin.

Some plugins have been written for Google Website Optimizer.  Many have become outdated without listed updates through current WordPress versions while other users have reported issues with usage.  The best approach is to find a solution that is easy to implement on an ongoing basis.

If you would rather not use a plugin set tied to Google Analytics, there is a good video tutorial series on lynda.com on how to use Google Website Optimizer.

Optimizely, WordPress Sites

Although this solution is NOT free, it does allow restructuring elements on the fly for multivariate split tests. It is also a more costly solution, depending on the number of visitors you wish to track within your tests (as of this writing, it can be as much as $359/mo. with larger traffic counts to use Optimizely) The more tech savvy will likely be more satisfied with the affordability of the A/B Theme Testing Plugin which is included in the WPMU DEV Membership.

Determine a Purpose and Get Running

(Establish a Purpose) (What Do You Want People to Do)

It’s always good to follow through to a purpose. Visits are good. SEO is great, but targeting actual follow through by the visitor to some trackable method really helps identify site hotspots and whether users will truly perform to some regard to your offering. A good simple start to begin the process is to split test a sidebar newsletter subscription against other elements of the site.

Conduct Split Tests
Unleash Fury with Some Intelligently Planned Tests

Getting Too Complex Adjusting Content and Elements

Comparing multiple elements against other multiple element changes is an easy cause of loss of focus on these tests. The idea is to determine which modifications cause the positive circumstance, so the best idea is to keep it as simple as possible and compare element by element.

Recap, Definitions

Traditionally references on the web are to A/B testing, split testing, bucket testing and multivariate testing.

  • A/B Testing, Split Testing, Bucket Testing: Most Commonly a single variable test samples compared against one another, if not very similar.
  • Multivariate Testing: A more comples form of A/B testing where more complex forms of multiple element changes can be tested against one another.

Why is WordPress a Good Platform for Conversion Testing?

WordPress themes are for the most part widgetized these days. This allows forms of testing without coding modifications, while some of those areas that aren’t widgetized can be transitioned to support widgets so these types of tests can be continued into the future utilizing the most obvious areas of change considering elements and hotspots. Its a good idea to become familiar with WordPress sidebars and different themes with A/B testing.

Natural Search Results vs. Pay Per Click (AdSense, etc.)

It’s much easier to target larger amounts of traffic patterned after keywords with Pay Per Click (PPC). Spending ad money without appropriate conversion is something that most everybody has done, and this type of testing can really leverage the ad budget, but testing on natural search traffic obviously makes sense as well given there is some conversion base the site is contributing to.

Testing a Site Over a Long Term

Tests can still be performed on an ongoing basis over time with small amounts of traffic, but this is generally not suggested as there are significant seasonal and behavior changes over time.  It’s best to have a site with some amount of stable traffic, promote heavily during the testing period or use Pay Per Click to enhance traffic for the testing period.  You really want to have a statistically significant amount of traffic for best results.

Again, see some detailed information on the split test plugin in action at WPMU.org

Read about some top SEO strategies from the last 3 years on a WPMU.org post.

Running Round Fire Fury Photo:  Editor B, Escalator Photo Featured on Homepage: hapticflapjack

Comments (4)

Participate