post

Easy Split Testing With Google Content Experiments And WordPress

If you’re interested in which headlines, offers, videos, prices, guarantees and copy increase the number of leads and sales your business gets, this post is for you. One warning though – you have to put into action what you learn here today. Once you finish reading this post, start your first content experiment. And if you’re one of those people that thinks A/B testing (also known as split testing) is hard, let me assure you it isn’t.

Everything you’ll see in this post is free. Here’s our ingredient list:

  1. WordPress
  2. Google Analytics
  3. Google Content Experiments (now in Google Analytics)
  4. Two Free WordPress Plugins

Now, let’s look at the exact steps you need to take to create this recipe and get the leads and sales you’re after.

For the purposes of this post, I’ll use a sales page as an example. If you want to learn more about how to create effective sales pages, join the Dempsey Marketing Tribe on Triberr to get exclusive access to an upcoming webinar on the topic.

Alright, enough selling on my part – here’s how to do this.

Step #1: Determine What You Want To Test

Use a compelling headlineOn any given sales page, there are literally a ton of factors you can test, including:

  1. Headlines
  2. Sub-heads
  3. Imagery
  4. Body copy
  5. Guarantees
  6. Forms
  7. Pricing

… just to name a few. The first thing you want to test though is your headline. A headline is super important as it’s the first thing everyone sees when they land on a page. You have about 2.3 seconds to get someone’s attention. That’s the job of the headline.

Any marketer worth his or her salt will have a swipe file of headlines. If it can be done, it probably has. Here are 16 different types of headlines you can create:

  1. Straight Benefit
  2. Offer
  3. Question
  4. Story
  5. News
  6. Reasons Why
  7. Testimonial
  8. Warning
  9. Dominant Emotion
  10. If…Then
  11. Item – Hype
  12. Finality
  13. List
  14. Curiosity
  15. Expert Positioning
  16. Extreme Value Proposition

Once you’ve decided on what to test, in this case your headline, it’s time to create some variations.

Step #2: Create The Two Variations

The key to knowing what it was that caused someone to convert is to test one thing at a time. There is something out there called multivariate testing, which is testing multiple factors at once, but that’s a bit beyond most and you need a lot of visitors to do it (think Google).

When creating your two variations create two versions of the thing you’re testing, and keep everything else exactly the same. If you’re testing a headline create two, and don’t change anything else on the page.

The easiest way to do this is to create your first variation in WordPress, and then use the Duplicate Post plugin to create an exact copy of the page. Then on the copy, change your headline and the URL and save it. Voila! Two exact versions with different headlines.

Step #3: Create A Thank You Page

Penguins always thank you for your purchaseEvery sales page has a call to action, typically some variation of “click here to buy my stuff.” Once the person makes a purchase they land (or should land) on a “thank you” page. This page is critical in measuring conversions.

A few tips for creating excellent thank you pages:

  1. Tell the person what happens now that they’ve ordered
  2. Provide links to your social media profiles and invite the person to join you there and say hello

Step #4: Install This Other Free WordPress Plugin

There are so many free plugins available for WordPress I can barely stand it. Here’s one you’ll want to install to help with your A/B testing, aptly named: Google Content Experiments.

Install and activate that little sucker and you’re almost ready to go. Something you may need to do is add a little line of code to the header file of your WordPress theme. The plugin author has directions here on how to do that. If you’re rockin a Genesis theme though, you’re good to go.

Step #5: Create A Goal In Google Analytics

We’re more than halfway through and you’re doing great! The next step is to create a goal in Google Analytics so we can measure our conversions.

When you log in to Google Analytics you’ll be taken to the list of website profiles you have in your account. Select the site you’re testing and then click on the “Admin” link in the top right.

Google Analytics Admin Link

Google Analytics Admin Link

On the Profiles tab you’ll see another tab that says “Goals”. That’s what you want. Create a new goal under one of your goal sets, and give it a name that relates to the outcome.

Google Analytics Goal Set Up

Google Analytics Goal Set Up

Now that we have our two variations, a thank you page, and our goal set up (and all the plugins to make this happen) it’s time to set up our experiment.

Step #6: Create A Content Experiment In Google Analytics

Google Content Experiments used to be a separate product – Google Website Optimizer. A few months back Google integrated it with Analytics and we now have a single location to go to for all of this. They’ve also made A/B testing extremely simple. Here’s how to set up an experiment.

If you’ve been following along with the directions above just click on the “Standard Reporting” tab at the top of the page. This will take you back to the main Analytics page. Once there, on the left-hand side of the page, select “Experiments” under the Content section.

Google Content Experiments In Google Analytics

Google Content Experiments In Google Analytics

Click on the “Create Experiment” button toward the top of the page to create a new one.

Enter in the URL of your first variation and click the big blue “Start Experimenting” button. Exciting!

On the next page, pictured below, fill out the details of your experiment.

Google Content Experiments Step 1

Google Content Experiments Step 1

When you’re finished click the “Next Step” button.

Now this is the cool part right here. On step 2 you can select a metric for your objective. This is a fancy way of saying to select the goal you created in Google Analytics earlier.

Google Content Experiments Select Goal

Google Content Experiments Select Goal

After you select the goal, I suggest you check the box beside “Rewrite variation URLs to original in Content reports” and click on the next step button.

Now we’re about to get back into WordPress. In step 3 you’ll need to add the experiment code to your original page. Remember that content experiments plugin you installed earlier? That’s going to come in handy now.

Select the option to add the experiment code yourself and copy the code. Now, go to the page for your first variation and scroll down until you see a widget that says, “Google Content Experiment Settings”.

Google Content Experiment Settings In WordPress

Google Content Experiment Settings In WordPress

Click the checkbox, enter your experiment code, and save the page.

Back in Google Analytics, click the “Next Step” button and you’ll see a box appear that verifies that Google found both pages and the code.

Now you run the experiment, and wait for the awesome to occur. Of course if you aren’t keen on waiting for serendipity to happen you can run a small ad campaign and send people to the page.

Step #7: Write Down Everything You Just Did

The last step is more of a suggestion – keep a testing journal. In addition to creating an annotation in Google Analytics of when I’ve started a test, I keep a spreadsheet that shows:

  • Experiment name
  • Element tested
  • Date started
  • Variation 1 url
  • Variation 2 url
  • Date ended
  • Variation 1 conversion %
  • Variation 2 conversion %

Once you have a clear winner – something Google will tell you – create another variation and test it against the winner.

ABT – Always Be Testing.

Corny? Yes. Highly effective in increasing the number of leads and sales you get? YES!

Step #8: Ask Me Questions

If you have any questions on this, ask away in the comments below.

Bottom Line: Test, Test, Test

As an entrepreneur it is your duty to test, test, test. Test everything – headlines, calls to action, pricing, guarantees. Everything.

With WordPress and Google Content Experiments (and this nifty guide) it’s very easy to set up A/B tests. Now go create your own!

Comments

  1. Hello,
    Great post. I have a question. I set up a split test for a squeeze page on WordPress. Code is all there and everything checks out ok. I don’t know if it’s actually working though, because the original page url should rotate out with the variation right? but it doesn’t. I only get the original. Any suggestion?
    Thanks, Jace

  2. The duplicate content issue is not a problem as long as you add a noindex tag to the page. The WordPress SEO plugin allows you to do this easily. You should also add this tag to your thank you pages.

  3. Hi Robert,
    Thank you for great post.
    Can you please explain how this works technically? Example I have site and have index.php page as main page.
    I create index1.php and index2.php and inject both with google experiment code.
    Do I need to make index1.php and index2.php public and replace with them my index.php?
    How visitors will get to those pages in case if I still have index.php as my live page and it is out of experiment? Or google does that somehow automatically and sends robots to those pages and not real visitors?

    Thank you in advance.

    • Hi Vas – can do.

      If you are using your main index.php page as the “A” version, then you’ll want to create an “index2.php” page as your “B” version. Add your Google Content Experiments code to the header of both.

      Then, as people come to your site, Google will automatically rotate the two pages for you. Some people will see version A, some version B.

      Google will keep rotating the two pages and collecting data for you until you stop the experiment. Then it’s up to you to put the winning page in place as you main index file.

  4. Hey Robert,

    Thanks for the post very insightful, I also have the same question as Gary:

    So I create homepage and homepage 2, but I don’t want google to spider homepage 2 or it be seen in my sitemap (using Yoasts WordPress seo plugin), how can you stop this happening?

    This become a real problem when you have 10 pages that you are split testing over time, you will get hit with duplicate content penalty.

    Cheers,

    Adam

    • So far I haven’t heard of anyone getting hit with a duplicate content penalty when they are using Google products for split testing. We do the same thing as you – use the WordPress SEO plugin to noindex the page and keep it out of the sitemap.

      • I would still be a little worried about duplicate content problems. Even with a no-index on the single page for the post it would still be included as part of other pages and feeds.

        With regards to doing it with pages as opposed to posts then I do not think there would be as many issues but with ordinary posts it would require a bit more work to prevent it from showing in other places, particularly the main blog feed page.

        However you may have hit onto something with the suggestion about when using the GExperiments tool the penalty doesn’t occur…

        Anyhow thanks for a good read, I’m gonna take note of your title type suggestions and agree with the conclusion completely: we always want to improve what we share and the only way to truly know how we are doing is test it. I spent a lot of time working within a multivariant layouts and flow campaign resulting is awesome improvements to engagement and CTR so the more testing you can do the better :)

        • I posted somewhat of a likely solution to what could be deemed as a duplicate content issue. I posted it as a reply to another comment farther down the thread but if your worried about the same dupe issues you should have a little search for the reply I left the other user.

  5. Very good post – I’ve been wading through a ton of nonsense about this stuff and glad to find this page…

  6. Gary Howell says:

    I’m trying to do split testing on my affiliate links. I’ve set the test goal to be an event. However the first time I ran the test I received no data. I believe that I need to add code to the affiliate link. Is that correct? if so what code would I add?

  7. Very nice tutrial

    But what i should do if I want to test Headlines on the main age. I want to check which headlines keep the longest time readers.

    And the second I would like to check the adense. How should i do it?

    • Hi Kamil – if you want to test the headline only then follow this tutorial to do that. You’d create two separate pages, with the only difference being the headline on the page.

      As for AdSense, we don’t use that so I can’t help you there.

  8. Hello

    Will this work for wordpress HOMEPAGE? So, basically showing 2, 3 or more different homepages?

  9. Kevin Bieri says:

    Awesome Post – was long on the search for this informations like the insertion of to the head.
    Greetings from Switzerland

  10. Hi Robert

    I’m having some trouble with getting the script to load on two different pages on a wordpress site. It also has Remarketing code on it and now I think that’s why I’m getting a validation error stating that Google thinks there’s broken code? Any suggestions?

  11. Hi Robert, thanks for taking the time to create this amazing write up!
    We all know you can create two different pages and have google show them to different people and measure which converts, but what do you do if you want to split test your pricing of particular categories in your online store?

    You wouldn’t be able to split test each product separately for each person because it wouldn’t make sense if someone found one product at price x in category a then found the same product in a different colour at price 1.4x in category a. I”m guessing a rule would need to be created to only track changes in category prices, rather than individual products and have g-analytics track that? Any ideas would be great.

    Thanks!

  12. Thank you for this nice tutorial. I almost though there was no website optimiser anymore. You prove me I was wrong ;)

  13. GaryHowell says:

    There’s something I don’t understand.  I know that Google will present one of two variations of a page when running an experiment.  But what I don’t get is that by publishing the test page on my wordpress site, won’t people, landing on the home page, see both the test and control page as separate posts?  How do I prevent the test page from being listed as a new post on the home page and in the categories?

    • GaryHowell hi Gary – this is for pages more than posts. I haven’t seen A/B testing used for posts, yet :)

      • Gary Howell says:

        HI Robert, Thanks for your response, however I tend to disagree with your assessment because in WordPress a page is nothing more than a post that doesn’t appear in the chronology of articles on the home URL. In my question I should have been more definitive by using the word post instead of page. To clarify, when split testing posts, how do I hide the duplicate version of the post from the homepage? And a further question would be that if I’m split testing affiliate links do I need to set up the content experiment as event tracking instead of destination URL since my destination URL is not on my site but on the advertiser’s?

        • That’s the major problem here. I said the same a moment ago in another comment. This is ideal for pages but requires some work if it’s to work well with posts. There is a way I have in my head that may or may not work, I havent tested it as I am going to create my own plugin for use specifically with posts and Google Experiments.

          I use a plugin on on of my sites to exclude aggregated content from appearing within the main feed and blog page but it also allows exclusion from archives and categories plus the search too if you wanted that. The plugin is called SimpleExclude and combine that with a plugin that can no-index, like WordPress SEO by Yoast, and exclude it from your sitemap and the post is essentially invisible.

  14. ToyinOmotoso says:

    “Install and activate that little sucker and you’re almost ready to go. Something you may need to do is add a little line of code to the header file of your WordPress theme. The plugin author has directions here on how to do that. If you’re rockin a Genesis theme though, you’re good to go.”The link you have for the instructions is Incorrect. It is supposed to be for the installation instructions at http://wordpress.org/extend/plugins/google-content-experiments/installation/Thanks for the tutorial and try to correct the link above so that it won’t confuse others.

  15. dude…this is some heady stuff. I really dislike doing A/B splits….Im not good at it, and it conflicts with my worldview that we’re all the same lol Amazingly useful post tho…you are a king, my friend :-)

    • dino_dogan split testing is one of those things that doesn’t get done as often as it should, mainly because it seems unapproachable. When Google integrated Website Optimizer into Analytics (brilliant move that helps them a lot) they made it a lot easier to do.Long story short – get to testing!

  16. christinejbrady says:

    Hi Robert,Awesome.  This post is just awesome.I, too, am a compulsive tester but I am not nearly as polished as what you describe here.  This is incredibly helpful information and with your step by step instruction, there should be lots of testing going on!I’ll be using this information wisely.Thanks for sharing!~Christine

    • christinejbrady hi Christine – glad it can help. Get to testing!

      • BarryCunningham says:

        RobertDempsey I’m having trouble trying to figure out where to add the experiments code to my POST (not PAGE). I’ve got my Genesis theme set-up to display a post as the home page but don’t know where the code goes. Any ideas?

      • BarryCunningham says:

        RobertDempsey I’m having trouble trying to figure out where to add the experiments code to my POST (not PAGE). I’ve got my Genesis theme set-up to display a post as the home page but don’t know where the code goes. Any ideas?

        • BarryCunningham hi Barry. As far as I can tell the post (or page) needs to be published before the “Content Experiments Code” section shows up. Are both of the posts you want to use for the A/B testing published?

Leave a Reply