Landing Page Optimization #3 (Testing & Analysis)

This is now the third in a series of three blog posts, providing a comprehensive overview of landing page optimization.  The first post discussed laid out the 5 tenants for effective online selling.  The second applied those concepts to design, marketing and implementation best practices. We are now ready to discuss how testing analysis form the basis of a on-going optimization methodology.

Even though there are known landing page optimization best-practices, and heuristic models for usability that guide us, testing is still at the heart of any effective optimization campaign.  It’s a process.  And the truth is that all of rules of thumb are just like any other advice in life – half true at best. At best you could argue that these rules are true some or most of the time.  No matter how great of an expert one is at site optimization, even the best minds in this space often find themselves surprised at what’s working, as evidenced by the success of sites like WhichTestWon.com, industry experts come to socialize and guess which landing page will win.

Thus what truly separates an expert from a novice is the systematic methods by which they arrive at an optimized page, more than anything else.  Think in terms of what a build > Measure > Learn feedback loop in which you build a landing page using best-practices to begin, then test it against a few other variants, and analyze the results to uncover what’s working. 

Based upon this effort, you not only determined which of your variants was stronger, hopefully you’ve revised your knowledge of what works for your target demographic, so that you can build upon it for the next batch of variants that you will test.  This process repeats until you’re convinced that you’ve optimized sufficiently, or that the law of diminishing returns catches up with you.

To setup your own testing process, you need to properly instrument your site with Analytics tools, and site optimization tools.  Let’s discuss each on of these:

Analytics
The first step is to properly instrument your site with Analytics, so that you can perform behavioral analysis after a test has run.  There are number of tools such as Google Analytics and Adobe Site Catalyst (aka Omniture) that will suffice and allow you to see who came to your page, what keywords they may have been searching how long they stayed, and where they went from there. You can pass tags also to identify campaigns and thus attribute conversions accordingly.  Goals can be defined as well that you can track against your attributed campaigns to uncover the cost of customer acquisition and return on ad spend (ROAS).

There are still other complimentary analytics tools that may make sense at times. For simple landing page optimization, you might also consider looking at CrazyEgg.com, as it provides a simple heat map to indicate where the mouse went on your page, thus indicating areas of focus.  If you’re looking at site wide and conversion optimization, a lifecycle analytics tool such as ClickTale  allows you to track (anonymously) the entire session on your sit, in detail.  Careful study can help you to uncover bottlenecks in your sales funnel that can be corrected for significant gains in profit and ROAS.

Always Be Testing
Separate from analytics tools, there is another class of tools that specifically manage systematic testing of content variations. Tools such as Google’s Website Optimizer (GWO),  Adobe’s Experience Management, Unbounce and others, enable you to  systematically test variations of a page and graduate the winner of a test to primary status.

A/B Testing provides a simple approach to compare different versions of a page whereas Multivariate Testing (MVT) testing variants of specific tagged elements within a page, such as swapping out a headline, image, or button.  Internet marketing companies that are really great at this stuff, make variant testing a regular part of their process and it is not uncommon to be running a dozen or more variants at any given time in these organizations. Internet marketing companies who excel at this stuff, are regularly testing as a matter of process, and it is not uncommon to be running a dozen or more variants at any given time.

Conclusion

So there you have it.  After three lengthy blog posts, you now have a fairly complete overview of landing page optimization.  At a very high level, the goals are to provide a simple and consistent user experience and to remove barriers from achieving our goal, such as opportunities for distraction and trust issues.  To execute on these goals, do research to better understand your demographic and the nuances of exactly what they’re looking for.  You can then respond with more effective marketing copy and improved design that takes advantage of layout for focus, and images to invoke emotion.

Be aware of implementation issues that could handicap your success such as load time, relevance, and SEO accessibility, particularly if dealing with Google AdWords.  And above all else, instrument your site with analytics to gain visibility into what “is* on your site, and make testing and incremental refinement a matter of process, to discover what *could be*.   If you can follow these steps and make a commitment to testing as a matter of process, then congratulations – you’re already well on your way to seeing double, possibly even triple digit gains in returns on ad spend.

Part 3 of a 3-part series on Landing Page Optimization

Website Optimization Tips

Optimizing a site can mean many things.  If you’re an eCommerce site, probably means optimizing your landing page, hiding your navigation so as not to distract, and pushing visitors toward the goal of a transaction receipt page.  For small business and service professionals, it may mean getting someone to fill out a form.  Or if you’re a publishing site, it very well might mean something nearly opposite of those two focused goals – perhaps you actually do want visitors to traverse your navigation, discover and engage with your content or community.  No matter how you describe it though, these things all have one thought in common:  you’re looking to maximize the value of each visitor to your site.

Although this practice is typically referred to as Website Optimization, Landing Page Optimization (LPO) or Conversion Rate Optimization (CRO), it is part of much broader schools of thought.  Many universities now offer interdisciplinary programs called Human Computer Interaction (HCI) that combine the study of Psychology and Information Systems, to better understand how to create systems that do a better job of servicing and assisting the humans that the systems were built for.  Usability and User Experience are two more common terms now used in design circles, to describe how to improve a user’s interaction with a system or website.  In many ways, landing page optimization and conversion optimization are applied marketing concepts, as much as Usability and User Experience are applied design concepts; but they’re all drawing up HCI which simply seeks to create better interfaces that accomplish the goals of the system. For the purposes of this blog post, I’m focusing on the marketing prerogative, but many of these thoughts can be applied to overall system design and usability as well.

Improving Efficacy
So how does one maximize value from their visitor? For the purposes of this discussion, let’s assume we’re talking about the easy to understand and quantify, eCommerce model.  With eCommerce, there’s typically a large spend on PPC and that spend is usually in pursuit of directly-quantifiable direct sales; not long-term fuzzier metric brand building. So our goal in this case would be to get users to our site for $1 and get as many of those visitors to buy something as possible, to increase average profits above the dollar we spent to acquire their attention.

Website Optimization
We’ll ignore the efficacy of the ads themselves for this conversion and focus purely on what happens after they arrive at the site.  Onsite, there are really two things that matter:

  • Focusing attention in the direction we want them to go.
  • Reinforcing trust and removing doubt and fear.
  • Keep attention focused and directed at the goal line (the receipt page).

If we keep those strategic goals in mind, the tactical implementation is relatively straight forward at a high level.  Obvious optimizations would be:

  • Remove the navigation bar on the landing page
  • Use very large submit buttons that are hard to miss
  • Use simple bullet points for details, not long-form text
  • Don’t present too many choices; match exactly the one product they searched for.
  • Use a headline that excites the user.
  • Use trust signals such as testimonials or credential logos on landing and checkout pages.
  • Minimize extra steps that could be hurdles to checkout such as the always-favorite required membership signup.
  • Remove barriers such the form that goes blank and requires refill if one piece of required information is wrong.
  • Don’t ask for more information checkout than is absolutely necessary.
  • For God sake, do NOT use Captchas!

Beyond those basics, there are a number of more subtle opportunities for optimization that simply require a “feel for the art” of it all, and proper domain knowledge, to match the culture of your users.  Catching their attention is key but authenticity and trust are also very important and cannot be sacrificed for a cheap thrill, or they’re not going to give you their credit card number.

Tactics that might work well would include using a larger checkout button with appropriate messaging.  Perhaps something exciting such as “Check it Out!” works for an entertainment product but a move conservative “Get Started” is more appropriate for insurance or financial institutions.  Color choices, images and titles work the same way.  Focus on finding a way to grab attention and infuse emotional excite, without compromising authenticity or trust.

Testing Methodology
As with any advice, there are many half truths in optimization.  What works well for many sites may not work for yours.  A lot of this comes down to culture and the issues f authenticity and trust I mentioned above.  For this reason, you may start with certain known principles as described above and broader usability heuristics, but the only way to truly know what works best for your own audience, is to test.

Testing in fact, is at the basis of most sophisticated online marketing operations today. We should start with a humility that we never absolutely know truth, we can only approximate it, and should always seek to evaluate truth through objective and empirical evaluation.  As such, most online markets today user an iterative approach to their online marketing and optimization strategies.  Continual testing and revision is baked into the on-going process.

To test your overall conversion funnel, you just need to setup goals and funnels in your analytics tools to be able to track and observe where bottlenecks are and results of changes. The efficacy of a specific landing page however, is better instrumented with a tool specifically designed to track changes of that one page, and track the goals against those changes.  There are tools such as Unbounce, Optimizely, and the free Google Website Optimizer (WSO) for that; WSO optimizer in particular is handy due to its one-click Google Analytics integration.

There are two types of tests that you might consider running for landing page optimization – an A/B test, or a Multivariate test.  An A/B test compares two or more separate pages, to determine which is more effective. This is a simple test to setup, requires less traffic, and is great for comparing entirely different versions of page layout.  If you instead want to focus on testing more subtle changes within a page such as title messaging, or button color, or featured image, a multivariate test would be appropriate.

A/B Testing
Using a tool such as Google Website Optimizer (WSO), you’d either setup redirection for the randomly served variants of yourA/B test, or you’d embed a few JavaScript hooks so that you can manage your variations from the WSO tool, if running a multivariate test.  Once setup, just wait for enough sample data to collect and let WSO do its thing.  Over time, you’ll see a percent-likely determination for which page is more successful. For example, “This page is 95% likely to outperform”.  Once you reach around 90-95%, WSO will determine the variant to be a “highly significant” improvement.  From there, you’ve confirmed which variant is more effective and can make your changes, as well as consider the next round of tests based on your findings.  This process can be repeated enlogica as you focus more and more on exactly what works.

For anyone new to optimization, I cannot stress enough, how important these methods are; you’re likely leaving a substantial amount of money on the table.  If you’ve never thought about optimization before, its quite likely that you can increase your Return on Ad Spend (ROAS) by 100% or more. I personally have had experiences in which small changes such as making a checkout but big, green and changing the messaging on it, with a single test, increased conversions by almost 40%.  So, while this form of methodical and detail-oriented optimization may not be the most sexy part of running your online business, it could easily be one of your most important!