5 Steps to SEO Friendly Split Testing, Sans Stress

This guest post is by Justine Smith of FreshBooks.

Split testing.

For many bloggers, the very words inspire visions of code and feelings of stress.

Also called A/B testing, split testing is an art that allows you to divvy up your blog visitors so one group sees a slightly different website than the other group.

While split testing for Adsense has been covered on ProBlogger before, doing split tests for your blog can be just as beneficial to help ensure that you’re giving your visitors exactly what they want.

To start you off on the right foot, here are my top 5 tips for SEO-friendly split testing that won’t damage your blog or stress you out.

1. Keep your test pages from being indexed

The first thing you need to address is keeping those test pages from getting indexed by the “Big G” (a.k.a. Google). There are several ways you can do this. It sounds a little techie, but I promise it’s painless.

First, use canonical tags. Using the rel=canonical tag lets search engines know where your “original” page is located.

Next, use your robots.txt file. A robots.txt file is created to provide critical information to search engine bots, and is one of the first things they will look for before they crawl a site.

With a robots.txt file, you can keep them from crawling as well as indexing certain pages, including your test pages. Just be aware that SEO-savvy competitors will often look at your robots.txt file to get ideas from your secret test pages, so be careful of what you include.

Use the noindex meta tag on your variation page. Again, this tells the Google Bot to back off.

And don’t forget to check your work. You can input site:yoursiteurl.com into Google search to verify that your test pages aren’t getting indexed. You can also use tools like Open Site Explorer or Majestic SEO to see if there are any external links pointing to your test pages. If there are, your site could get indexed—so be sure to check your work.

2. JavaScript doesn’t always work the way it’s supposed to

JavaScript is a programming language that’s used to make web pages interactive.  In the context of a split test, it is a piece of code you place on a web page that enables you to actually ‘split’ traffic between your control page and your testing page.

Search engines aren’t supposed to see or index Javascript, and usually they don’t.

To split traffic between pages A and B, JavaScript is often used at the top of the control page. Common sense tells us that if search engines aren’t supposed to follow the JavaScript then they shouldn’t index the test variation URL.

Unfortunately, sometimes they do it anyway.

So once again, check your work. Do a site: search for your test URL every so often to verify the test pages aren’t being indexed. You can also segment your site analytics by search referrals and note the top entry pages as another way to see if any of your test URLs are getting indexed by accident.

3. Take steps if your test pages are indexed

Sign up for Google Webmaster Tools. Besides the option to remove your accidentally indexed pages, this toolset gives you a plethora of other free tools that’ll get your site much more search visibility.

Also, be mindful of:

• Keyword cannibalization: This is where two similar pages wind up competing for the same rankings. Ultimately Google will arbitrarily favor one and ignore the other. There are several rank-boosting benefits that can be lost when this happens, so do your best to avoid it.
• Duplicate content: This is another biggie that can have similar results to cannibalization. If you don’t keep careful tabs on this, test URLs that are indexed could result in your important pages getting de-indexed. Not exactly what you were hoping for, right?

4. Try testing SEO-safe page elements

This is a case where small changes can add up to big gains. Did you know a simple color tweak could make a big difference on your conversation rate? The bonus is these changes won’t affect your rankings in any way.

Consider testing:

• Headlines and calls to action: Simple changes like text size or typeface can be a great way to shake things up and impact conversion rates. Different people respond to different things, so making changes will help you to gauge what your readers respond to best. Just be sure when changing text size not to use the header H1-6 HTML tags, as those are reviewed by Google.
• Colors: Another great way to split test without messing with SEO is to play with colors on your site. Once you get a good look at all the areas where color lives on your landing pages, the possibilities will seem endless. Speaking of color…
• Text color: This is also something that can be tweaked. Changing a few important words to another color can make a difference. Just remember not to use too much color, because then it just becomes distracting and folks are put off. You can also change highlighted text, both by changing the color of the highlights and also changing which words are highlighted.
• Adding and removing graphics: Finally, manipulating your graphics can also effect conversions. Simple changes like removing graphics from your site altogether are easy to implement. Google does index photos but it’s not going to break your SEO if you no longer have them on your site. By the same token, you can also add photos to spark a change. People respond to social proof, so adding images like press logos and testimonials from people who adore you can go a long way in increasing response rates of your visitors. Just be sure not to mess with your existing photos’ ALT text too much, since this might already have been indexed.
• Changing graphics: These can be simple changes. You can swap out an illustrated graphic for a photograph, or see if people respond better to stock photos or real-life snapshots. Or you can play around with the positioning and size of your images. If the thought of changing your graphics makes you queasy, you could try simply playing around with the things that compliment your graphics, such as using borders, overlays and text on top, just to see what happens. Oh, and tying in with the call to action tweaks from earlier, you can also tweak that Buy Now graphic you may be using. Rumor has it The Belcher Button with its orange hue, and credit card images make a huge impact on buyer numbers.

5. Give Content Experiments from Google Analytics a whirl

As the “artist formerly known as Google Website Optimizer”, Content Experiments might be of interest in your split testing fun.

This is basically a quick and relatively easy tool that you can use to create A/B tests without being overly technical or messing with a lot of code. You can test your main pages and even get email updates about how your experiment is faring.

A final word

Split testing is an essential long-term tactic for a variety of business types, including freelance bloggers and even bigger companies like FreshBooks, who use the results to make educated decisions about how to write and design for potential and existing customers.

Now that you’ve ventured over to the geek side, it’s easy to see that split testing isn’t nearly as scary as many people tend to think. As long as you know where some of the hiccups lie, there’s no hair pulling, code whizzing, or messing with your hard-earned rank required.

Justine Smith is the Outreach Manager at FreshBooks, the #1 Cloud Accounting Specialist for Small Business Owners.