Split Testing | Improve Results Without Spending More Money

by Jennifer Horowitz on September 10, 2012

Today it’s all about improving results. One of the best ways to improve your results is with split testing (multi-variate testing works also, but I’m going to keep it simple today and just talk split testing).

Testing is vital to the success of your website; it is a powerful strategy for increasing the effectiveness of your site pages.

Split testing is also known as A/B testing.

So for the beginners out there, it is the process of taking one page, or ad (your control or page/ad A) and then creating another page/ad (page/ad B) – not you only make one change on the B version. You use software to split the traffic equally among both versions and then you track the results. One page/ad will get better results (usually and if it doesn’t, you need to test something else, probably something more drastic). Once you have a winner, that becomes the new control to work with and you create another version to test. With continual testing, you can improve your results and get more out of the advertising/SEO/social media dollars and time you are spending to drive traffic.

You can split test landing pages, ad copy, headlines.

I love split testing. It takes opinions and personal preferences and guess work out of your marketing. I also love it when testing proves me wrong (when I think something will win and it doesn’t) – it’s a great way to learn and it improves my results.

Think about it, if I were to just go with what I had thought would win, I would be constantly getting lower results than I can now get with my new page which has proven to pull better results.

We just don’t know how the market is going to respond and even with proven marketing concepts in place and great copy, we just can’t predict the response. With testing, we don’t need to predict. We can accurately and scientifically determine what is going to work and improve our results.

Think about it, you are running various campaigns and driving traffic to various pages anyway. Why not get the most out of those pages and increase your response without increasing your expenses?

Here is a quick breakdown of how to split test:

Step One: Identify a page with an offer on your website that you would like to test.

Step Two: Create another variation of that same offer (in order for an A/B test to be accurate you can only change one thing on the page)

Step Three: Place your split testing software code on your site, and drive traffic to the pages.

Step Four: Review the results and determine which page was the winner. That page is your new control page, and you can now create a new page B to test something else. The idea is you keep testing and keep improving the results.

You can test anything, but here are some of the most common things to test:

Pricing (example: $9.95 versus $12.95)
Headline color (example red versus blue)
Headline text
Header graphic versus no header graphic
Negative versus positive text (Ready To Fall In Love? – that’s a positive approach. Sick Of Being Alone? – that’s a negative approach)
Different images
Different font sizes
Different calls to action
Buttons versus text for your calls to action
Weird images (like you see on all those Facebook ads) or “normal” images
Smiling versus unsmiling people (yes, something that small can impact your results!)
All text versus text and images
Different layouts
Professional stock photos versus more amateur looking “real” photos
Border versus no border
Background colors and images

The list is endless and it depends if you are testing an ad or a landing page.

If you aren’t currently split testing, I guarantee you aren’t getting the maximum benefit out of your website and you need to start!

Here are some examples of split tests:

Example One: PPC ad:

Version One:

Designer Shoes are Affordable
Women’s Designer Shoes
Big names, great prices

Now, let’s say your click through rate (number of people that click on your ad) is 4.2%. You are happy, because that’s better than you’ve ever done before. What most people do is keep running with that ad. What we want you to remember to do is Split Test.

All that means is create a second version of your ad, to compete against your first version and see who wins.

So, now look at version 2:

Designer Shoes are Affordable
Lady’s Designer Shoes
Big names, great prices

Do you see the difference? Only one word (more about that in a minute). In the first ad we have

Women’s Designer Shoes and in the second ad we have Lady’s Designer Shoes.

So, let’s look at the results.

Ad 1 – 4.2 click through rate
Ad 2 – 5.1 click through rate!

You just increased your response rate by 21.4%! No extra costs, simply changing one word and getting an improved response.

Why did more people click on Lady’s than Women’s? We don’t know, we may never know, but we also don’t care. All we know is have now improved our click through rate!

Now are we done? Not even close. Now we take Ad 2, and create another ad to split test against Ad 2. You can constantly test 2 ads against each other and incrementally increase your response rates.

Example two: Landing Page Headline

Version One:

Are You Ready To Learn Everything You’ve Ever Wanted To Know About SEO?

Version Two:

Everything You’ve Ever Wanted To Know About SEO

In version one, I posed my headline as a question, in version two, it was a statement.

Results: The question won – by a small margin, but still a winner.

Example three: Order Buttons

Version One:

Version Two:

Results: Version two won decisively!

Example three: Stock photos

Version One:

Version Two:

Results: Version two won.

So you get the idea, nothing is too small to test. Look at your site, are there big pages for you (ones that get a lot of traffic and are important for your business)? Try setting up a split test on those pages. Try to improve your results.

Are you doing any paid ads on Google or Facebook or anywhere else? You should be split testing not only the ads but the landing pages you drive them to.

If you are doing email marketing, many of the Autoresponder programs allow you to split test subject lines.

There are tons of options out there as far as software. Some are free, some are paid. Google used to offer Website Optimizer but they just phased it out. They have a replacement tool that you can use but I haven’t tried it yet, I have a paid tool that integrates with my stats program. Just do a little googling and you’ll find many options when it comes to software.

Let’s have some fun – if you want to see what I would split test for you, post a page or ad text and I’ll come up with a couple different things for you to test. :)

If you have experience with split testing, I’d like to hear your thoughts and what things you test!

Jennifer Horowitz

With over 13 years in the industry, Jennifer Horowitz, Director of Marketing for EcomBuffet, has amassed much knowledge and experience and has much to say about all things SEO (marketing, copywriting & social media). Always happy to share with an audience, Jenn is now a regular contributor at Level343.