4 Lessons for Testing Landing Pages

Jan 14, 2013 9:50 AM  By

If you’re not testing your top trafficked site pages, you’re losing competitive advantage, revenue optimization, and sizeable UX gains. One loss is enough to dishearten you; a loss in all three drives proverbial nails into your ecommerce coffin.

Here are some lessons I’ve learned from recent testing from 700,000+ visitors across dozens of landing pages.

An iterative approach rules
By iterative, I mean leading internal staff and stakeholders to embrace the strategy of “beat my control.” To do so, you have to be laser fast at front-end creative development and even faster identifying which elements you want to test. In other words, know what you want to test (more on this later). Attract and employ great people who buy into the mindset of being fanatical about optimizing and never settling. This requires folks to strangle their egos daily. It also requires a fast pace and environment. This attitude should permeate your ecommerce DNA.

Results vary depending on traffic type
Realize that most ecommerce traffic today is channel agnostic. Visitors research, purchase, return and interact via different devices. For example, when driven to your site via PPC, they will convert differently than when driven by affiliates or third party email rental or SEO.

I can think of one example where an affiliate relationship was driving traffic to us from email. Their lists were pre-qualified and converted differently than cold traffic we were driving via PPC. Understanding that traffic types—and the channel used for driving that traffic—impacts conversions. This can mean your current landing page control may sadly underperform or even lose big to new tests where the your inbound traffic driver is different.

There are always sub-tests lurking beneath your tests
Your test may be a new landing page design versus a proven control, commonly called an A/B test. But don’t forget the minutia. Things like dynamically serving headlines, showing varying forms or shapes or colors of order buttons, or displaying/suppressing certain opt-in fields are all prime examples of common sub-tests living under your larger test.

Tested all together, and this is called multivariate testing. So don’t ignore this low hanging multivariate fruit. I’ve experienced majoring in these minor details can bring indeed large lifts.

Be careful what you test and chase
Everything can be tested, but that doesn’t mean it should be! A marketer’s intuition and a healthy curiosity come into play here. The key is to have a curiosity to offer large lift hypotheses, and test accordingly. Otherwise, you’ll drive yourself crazy chasing small issues that can have marginal impact.

There’s only so many hours in the day, and having a healthy dose of questioning of ‘WHAT-IF’ is certainly fine. Just don’t chase everything or you’ll burn out yourself and your staff. Stay narrow on what tests you want to deploy that can add large incremental growth and breakthroughs.

I said earlier you must strangle your ego daily. The same can be said of your staff and executives. With any form of testing, data rules. You (and they) are never the best judge of opinions or conclusions of the greatness of your front-end marketing. You are too close to your company and its products and services to be able to objectively determine the effectiveness of your front-end marketing without testing.

In the long run, testing is the only true way of collecting data points on your website and pages, and in forming conclusions on their effectiveness. Remember the results of your test experiments represent what site users are expressing.

Daryl Logullo is the director of ecommerce for American Lantern Press, which publishes news, view, and analysis to empower savvy, self-sufficiency seeking individuals in protecting their assets, health, freedom, and privacy. ALP is parent to IndependentLivingNews.com and IndependentLivingBullion.com.