Conducting a Google Content Experiment with SEO in Mind

How to use Google Content Experiments to improve conversion rates, without affecting your SEO performance.

A/B testing (often called split testing) is the technique of comparing the performance of two versions of a web page to see which one shows the best results. Generally your traffic is split between these two versions, over the same time period, to generate statistics for each version.

‘Google Content Experiments’ is a fantastic (and free!) tool provided by Google to allow webmasters and online marketers to conduct A/B and multivariate testing on their websites, which feeds right into your Google Analytics Dashboard.  However, any form of split testing relies on creating various versions of a single page or on-page element.  Where we have more than one version of a page, we have duplicate content.

Although Google has directly stated that a little duplicate content won’t hurt your rankings, we certainly don’t want more than one version of our home page indexed!  With all this in mind, the following is a guide on how to conduct a Google Content Experiment in an SEO friendly way:

Step 1: Preparation

Goal setting has always been an important part of traditional marketing; the same goes for digital marketing.  Before we look at the how, we have to decide what we’re trying to improve with our content experiment.  Do you want to:

  • Increase conversions?
  • Reduce bounce rate?
  • Increase page views?
  • Increase shares or comments?

The list is essentially endless and depends entirely on your website and your business objectives.  Let’s assume in this example that you want more of your visitors to sign-up to your newsletter.   If you don’t already have sign-ups set up as a Goal in Google Analytics you should do so.  If you’re not comfortable with Goals, Google provides a great guide.

“#Google have directly stated that a little duplicate content won’t hurt your rankings”

Step 2: Create your variations (don’t forget about SEO!)

Next you’ll need to create the variations of your page (you may need to use your web designer / developer for this stage, depending on your skill level).  At this stage, you will need to decide whether you’ll want to test two versions of a web page (A/B test) or many elements within those pages (Multivariate test) – this guide from Hubspot offers some practical advice on each.  For the purposes of split testing, there are a huge number of variables which can play a part.  These can range from your sales copy to the colour of a headline, but below are a few ideas:

  • Headlines – Are they eye catching?
  • Sub headlines – Do they add additional value?
  • Paragraph Text – Is there too much text?  Is your copy persuasive?
  • Testimonials – These add an element of trust, but are they convincing?
  • Call to Action text – learn more by reading ‘What to call your action
  • Call to Action Button – Is this eye-catching enoug to mximise conversions?
  • Links – Do you have too many links? These can act as distractions
  • Images – Good images send all the right messages
  • Content near the fold – Below the fold content may not receive as much attention

In our example we’re trying to improve our sign-up rate, so perhaps we’d create two version of our page with variations in the style and location of our call to action. You may want to read our guide on increasing conversion rates with targeted landing pages.    When it comes to implementing your pages, they should each be uploaded on separate URLs.  If we’re testing our home page, they may look like this:

  • http://www.example.com (control)
  • http://www.example.com/variation1 (variation 1, required)
  • http://www.example.com/variation2 (variation 2, optional)

In order to prevent duplicate content issues, or to stop Google indexing our test pages, we have two options.

  1. Noindex the variation pages in your Robots.txt file or on-page.  This is a universally recognised signal which requests that search engines do not index the variation page.  However, if for whatever reason you generate inbound links to these variation pages, the value of those links may be lost.
  2. As such, the better option from an SEO perspective may be to add Canonical elements to your pages to nominate the preferred version, thus retaining any link equity.

No matter which option you choose they act as a temporary measure as, following the completion of the experiment, you should permanently (301) redirect all variation pages to the final version.

“If you don’t already have sign-ups set up as a Goal in #Google #Analytics you should do so.”

Step 3: Set up your experiment in Google Analytics

1. Log into your Analytics account and navigate to Reporting >> Behaviour >> Experiments 
2. Enter the URL of the page which you’re looking to improve and Start Experimenting!
3. Name your experiment
4. Select your Goal or metric
5. Choose which percentage of traffic you would like to send to your experiment

Content Experiments

6. Add your control page (normal home page)
7. Add your variation page and off you go!

Variation Pages

 

Step 4: Review and improve

You’ll now be able to monitor the results of each of your variation pages versus the original, over time.   The experiment can run for a minimum of three days and up to a maximum of three months.   If your Google Experiment reveals one page which clearly performs better than the others then the obvious response is to replace the original page with the variation which performs best.

However, it’s more often the case that there is no obvious winner.  In this case, you may want to revise your experiment or launch an entirely new one.  One common mistake is making the variation so small that there is no discernable difference in performance.  In these cases, try making some more substantial changes and re-test.

Remember, following your experiment you should 301 redirect the variation pages to your designated page.  Good luck!

P.S.  If you’d like to know more about Google Content Experiments, the below video provides further information and instructions:

We're recognised by
02034321369
Digital Marketing Agency London +Mike Grindy