Last Month’s Earnings

$213,212.64

Currently Reading

How to Run Free Split Tests Using Google Optimizer for Better Results

How to Run Free Split Tests Using Google Optimizer for Better Results

By Pat Flynn on

Do you want a poor website? Of course not. We all want to make our websites the best websites they can be. We all want them to grab more attention, catch more subscribers, and increase our sales and popularity.

In a previous post, I talked about how I spent just 5 minutes to change my “buy it now” button, which this increased my book sales by almost 25%. The new button I used was proven to work because it was “split tested” with a multitude of other buttons with different sizes, shapes and color combos.

To recap, a split test is a test that involves showing different variations of something to equal amounts of different and random people. The test will determine which variation does better than the others.

Like I said before, companies will spend millions of dollars to do these types of tests using a number of focus groups and analysts – it’s ridiculous. Thanks to Google and the power of the internet, we can do this with ease for free.

Things You Can Run Split Tests With

Before I go over how to run split tests yourself, you may want to give some thought into what you may want to test. Here is a list of different things you can create variations of for split testing:

  • Colors: including background color, font color, and button and graphic colors.
  • Sizes: including font sizes, graphic sizes, button sizes, page sizes
  • Location of Objects: for instance, placing an action button above the fold vs. below the fold of your web page. An action button is something like a link, email submit, or buy button.
  • Order of Objects: such as testimonials, features, price, advertisements, examples, descriptions, etc.
  • Graphics: including jpegs, headers, buttons, etc.
  • Content: meaning the text that is on your page. This includes the headings, the intro, the body, the conclusion, testimonials, bonuses, advertisements, etc.
  • Prices, offers and guarantees

There are thousands of things to test, but it’s up to you to find out which things work best for you and your website.

Things You Can Test For

Using different variations of the things in the list above, you can look to improve your conversion rates on the following things:

  • Sales: The number of products you have sold with each variation
  • Subscribers: both for Email and RSS feeds
  • Email Submits: for your newsletters or tips using an email service such as Aweber
  • Clicks: especially for those affiliate links that earn you cash

How to Run Free Split Tests

First, you’ll need to set up Google Website Optimizer. Don’t worry – it’s totally free. If you have a user name for Google already, like for GMail, you’re good to go.

You can visit the Google Website Optimizer tool here.

After a brief setup, you’ll end up on your experiment list page, which is basically your admin panel for your tests. This is what it looks like (click to enlarge):

In order to begin, you’ll need to click on create a new experiment. Then, click on A/B Experiment (there is another form of testing called multivariate testing, which I will go into at a later date).

There are three things you need in order to run your experiments:

  1. An original page that you are running the test on. You’ll need the url of this, for example: http://www.yourwebsite.com/index.html
  2. A variation page that is just like your original, which has the one difference you’re trying to test. This should be a totally separate web page on your server. You can just create a new page, copy paste the code from your original, and insert your variation. Your url may end up looking like this: http://www.yourwebsite.com/index2.html
  3. A conversion page, which is where your users will end up after doing an action on your page. For example, a thank you page: http://www.yourwebsite.com/thankyou.html

You can create more than one variation, but you will need separate urls for each.

Google will generate scripts that you must insert into each of the aforementioned web pages. This is so that Google will be able to distribute your readers evenly to each variation, so you can get accurate test results. This can be a pain, but luckily for those of us using WordPress, there’s a plugin.

The plugin is called the Easy Google Optimizer plugin from Impression Engineers, and is free to download from their website. Load it as you would any other WordPress plugin, follow the instructions, and you’ll have your test up and running in no time. Many thanks to Impression Engineers for this awesome tool.

Note that new visitors to your will be shown only one variation of the experiment, even if they were to come back and visit later. This holds true until you stop your experiment, which can be done at any time.

Give the test some time, and after a good amount of visitors, you’ll be able to see the results. Your reports will show you the % chance to beat the original, the observed improvement, as well as the number of visits and conversions for each variation.

After you know exactly which variation work best, stop the experiment and make sure that version is the one all of your visitors will see from now on. Then, rinse and repeat with another experiment. It will take some time, but after a few experiments, you’ll have a website that is optimized to convert as much as possible.

My Own Split Tests

Since this post is already lengthly, I’ll go over a split test I did myself and my exact results in the next post.

I hope this information has helped you and you begin to think about split testing on your own websites.

Thanks again for your support and your comments!

Email address required
No thanks, I'll pass for now :)
  1. 1. Free Stuff

    You'll get instant access to free resources, including my most popular book, Ebooks the Smart Way! (Downloaded over 125,000 times!)

  2. 2. Content Tailored to You

    Over time, I'll get to learn more about you and deliver content that actually matters.

  3. 3. No Hype

    Just real content that's meant to make a difference.