By Mark Sherbin published September 23, 2013

A Guide to Content Optimization with Performance Testing

man testing pages online-content optimizationHow can you know if your content marketing is pushing the right buttons if you aren’t testing it?

You can’t. Without testing, you’re missing a vital frame of reference — how one piece of content performs against others. This key piece of information helps you better understand which of your content pieces are hitting the right mark, and where you need to focus your content optimization efforts in order to increase your content marketing success.

Performance testing

Performance testing pits multiple pieces of content against each other under the same circumstances of timing and placement.

A/B or “split” testing is a mainstay of digital marketing. It’s a randomized experiment where two possible version options — i.e., two web pages, two design strategies, two pieces of content, etc. — are presented in equal scale to different viewers. Perhaps you’ve used the technique yourself to test the efficacy of your landing pages, or other elements of your digital presence.

Say you wanted to test the same landing page with two different headlines, for example. A/B testing processes randomize the headline that your visitors see — i.e., when they reach your landing page, some will see Headline A and others will see Headline B. From their reactions (i.e., did they spend additional time engaging in the content on your page, did they exit immediately, did they move on from the landing page to other pages of your site, etc.), you can get a snapshot of which headline option is more likely to produce the results you desire — a key piece of information for guiding content optimization decisions.

Yet, A/B testing is just one technique for experimentation. If you want to test several unique elements of your content marketing at the same time, multivariate testing (MVT) provides a systematic way to gauge the relative value of each available option.

The additional insights offered by content testing techniques — and the content management technologies that enable them — strengthen your ability to target the right customers. For this reason, as well as others discussed below, it’s critical that content marketers use testing to identify potential optimization opportunities and advantages for targeting and engaging audiences.

Four top contenders for content marketing testing

Ramona Meyer-Piagentini, Senior Consulting Manager in Digital Marketing for Adobe, asserts that testing is a cornerstone of today’s marketing, no matter where your areas of specialization lie. Her primary focus is on Adobe Target, the company’s revamped testing and targeting platform.

It’s one thing for an organization to have a content delivery model. But it’s an entirely different thing to execute well in the space,” she explains. “Intuition doesn’t play a role in marketing. If you’re going on complete assumptions, you could be missing the boat on driving real impact.

Once you’ve accepted that testing plays an essential role in any content marketing program, the next step is to start with some simple A/B testing — particularly if you don’t have a strong experimentation platform already in place.

For content marketers, A/B testing has at least four major applications:

1. Layout and design have been A/B hotspots since the dawn of the website. Tweaking small design elements can have a major impact on how your audience interacts with the page. 

For instance, content marketers may want to play with page elements like “suggested posts” on a specific piece of blog content, or decide between showcasing a video and an eBook at the top of their content hub’s home page. In fact, SAP used A/B testing to examine whether using thumbnail pictures would increase interest in its company news spotlight. The testing showed that the addition of thumbnails and clearer calls to action resulted in a 62 percent lift in engagement. 

2. Even small tweaks to headlines, opening paragraphs, featured pictures, and other elements of a piece of content can produce a powerful impact on that content’s success. Testing these components is a great way to fine-tune your writing and direct your attention to the areas where additional content optimization might be merited.

Take headlines for example — the component that accounts for much of the first impression your audience gets of your content. Running two different headlines in an A/B test can help you understand which headline styles are more appealing to your target readers and, therefore, are likely to have a greater impact on the value they perceive.

You can even get as granular in your testing as changing a link in the text to point to one of two distinct product pages. Whichever product page performs better in terms of engagement or encouraging action can help you set a baseline determination of the best content elements to use when targeting that audience segment.

3. Mapping content to your sales funnel can be a tricky task. Without experimentation, it’s tough to understand which content works best for leads in the top, middle, and bottom of the funnel.

Content marketers can perform A/B testing for purchase intent in a variety of ways. For example, pretend you have a group of leads that signed up for your newsletter on the same day. Now, deliver two separate newsletters highlighting distinct pieces of content — one focused on industry thought leadership and one focused on a question specific to the purchase decision.

If the purchase decision content performs better, you gain more insight into the meaning behind a newsletter sign-up — specifically that your audience usually signs up for your newsletter when they’re ready to buy.

4. Understanding your social media audiences can also be a complicated task. Some social channels have very clear demographic splits — for example, women are five times more likely to use Pinterest than men.

Audience preferences can be even more obscure when making comparisons between social channels, which makes testing more critical for determining things like segment distinctions between SlideShare and Twitter. Delivering different forms of content through A/B testing is a great way to learn how your content might perform on unique channels.

Using the scientific method for testing

Jordan Johnson, Manager of Digital Optimization for Southwest Airlines, approaches his job of building promotional content for the airline’s website with a scientific bent.

Testing and targeting needs a holistic approach,” Jordan explains. “It’s more than just understanding whether a red or blue button works better. It’s testing hypotheses to find theories about the effectiveness of your content.”

Jordan applies scientific method when experimenting with different forms of promotional content on the Southwest site. The same approach he uses on announcements, deals, and other brand-centric content can be applied to your content marketing:

  • Pose a question: For content marketers, this could be as simple as asking, “How can I improve the bounce rate on my blog?” This question could have several answers. Your next step is to propose one based on knowledge you already have.
  • Create a hypothesis and a prediction: Your hypothesis in this case might be: “The placement of ‘suggested content’ on the page influences whether or not visitors click on it.” The prediction in this case might be: “Moving ‘suggested content’ to a more clearly defined space on the page will help reduce the bounce rate and increase the average pages viewed per visit.”
  • Test your hypothesis: Here’s where you design your study. For this example, we’ll use A/B testing to test our hypothesis. Let’s move ‘suggested content’ from the right sidebar to the area just below the top post on a blog page. Using the right tools, you would then run your test and collect the resulting data.
  • Analyze the data: Now that you have your data, you can look at the impact of moving ‘suggested content’ to a new area. If the move reduces the bounce rate and increases the average pages per visit, your test can be deemed successful. But pay close attention to other factors that may also be impacted by the action you take to test your prediction. Did your conversion percentage drop as a result of this change? You may need to do an additional test to find out.

According to Jordan, a holistic approach to testing helps marketers keep the big picture in mind. “If someone comes to me asking to test a red button against a blue button, I can record the traffic data. But if they don’t share KPIs with me, I’m missing the perspective I need to help them frame those numbers.

Whatever factors you choose to test, make sure you have a strong frame of reference and a good idea of the ultimate goals you’re trying to accomplish before designing your test case.

Five steps for creating a content experiment

Testing takes careful observation and a strong pool of results to truly understand how the small changes you make might ultimately impact your overall content optimization efforts, as well as specific pieces of content.

Ready to get started with your own content experiments? Here are five steps to creating a campaign.

  1. In most cases, your control is the version of a piece of content or design element you already have in place or would usually choose as your centerpiece.
  1. You need a benchmark to understand performance. What are you trying to accomplish with the piece of content you want to test? For example, do you want readers to move on to another piece of content, visit a product page, or sign up for your newsletter?
  1. If you’re planning a straight A/B test, you’ll only need a single variant to test against the control. For more complex experiments, you may need to use MVT to compare multiple variables against your control.
  1. (for A/B or A/B/n testing): Changing multiple page elements makes it nearly impossible to identify exactly which of the design or content changes you made spurred the desired behavior. Decide what you want to change — the headline, featured content, picture, etc. — and only change that item.
  1. Collect an equal amount of visitors for each variant. When you have a significant sample, you can compare the results of your tests. A true A/B test requires 95 percent confidence in the results. (Here’s a helpful post on figuring out what that means for your test.)

Tools you can use for testing

More advanced content management platforms like the Adobe Marketing Cloud and HubSpot come equipped with tools to help you test your content. If your CMS doesn’t have this advanced functionality, you’ll need to look beyond the walls of your software.

Check out some additional options for creating your own content experiments:

  • Proprietary software platforms like Optimizely, Visual Website Optimizer, and Unbounce all represent great resources marketers can add to their websites simply and efficiently.
  • Content Experiments is a relatively new feature of Google Analytics that takes A/B testing to the next level. It’s specifically rooted in “A/B/N” testing, or a more complex environment that lets you test up to five different versions of a page.

How are you testing content components?

If you have other strategies for testing content optimization options, we’d love for you to share your experience with us in the comments.

Looking for more guidance on measuring and optimizing the success potential of your content marketing? Read CMI’s eGuide on Measuring Content Marketing Success. 

Cover image via Bigstock

Author: Mark Sherbin

Mark Sherbin is a freelance writer specializing in technology and content marketing. He shares occasionally insightful information at Copywriting Is Dead, where he promotes authentic communication between organizations and their audiences. Contact him at msherbin@gmail.com.

Other posts by Mark Sherbin

  • http://www.influenceandco.com/ Matt Kamp

    Excellent post, Mark. Great takeaways and unique spin: tackling a relatively subjective area (the ROI of content) and applying a scientific process to it.

    • Mark Sherbin

      Thanks Matt!