Skip to main content
ab-testing-1
18 February 2016

CRO Best Practices: A/B Testing part 1

Author

After our first 2 articles on Personalization in our series on CRo Best-Practices(here and here), let’s take a look at our A/B Testing recommendations.

Preamble: When should you favor A/B Testing?

A/B Testing is most often used to improve the global user experience (visual hierarchy, information architecture, ergonomics, …) while Web Personalization is used mainly on the value proposition of your offer (contextually relevant content and messages for specific visitor segments).

Whether or not you have conducted a deep analysis of your website, A/B Testing will allow you to determine if any changes you want to apply actually improve your conversions or not.

A/B Testing is not a silver bullet

We always recommend clients to start with A/B tests on cosmetic changes (wording / style on CTAs, new taglines, …), because they are simple and take you through the motions.

But there are NO magic color, layout, wording. Don’t take articles telling you “We had a 287% increase in signups by changing the color of our CTA to pink” literally.

What’s important there is not the color itself, but the visual hierarchy in the page.

Though you can use best practices and proven models as starting points, what worked for others won’t—in most cases, work for you. Every business / audience is unique.

“Quick wins” like these might get you going, but they won’t take you very far.

In fact, when it comes to conversion optimization, you won’t struck gold by focusing on a single element on a specific page, or an intuition or by copying a successful site.

It’s a deliberate process, based on data and a habit to insert in your decision-making cycle.

The following method will set you in the right direction:

Before you start launching tests left and right, let’s take a deep breath and start with a high-level view:

  • What is my website strategy?
  • What are my short-term goals? mid-term?
  • How do I measure success? What are my KPIs?

It may seem obvious but writing these down will allow you to align your tests with your overall strategy, get granular on your website activity and stay focused on what’s important for your business.

2. Know your market perfectly

You need to know your market inside out, what your competitors do, their positioning etc… Listen closely to what experts, influencers, and people say on forums, social media, comparison websites about you, your market and your competition.

Tools like Google Alerts, Twitter Search and Yahoo! Answers can help you get a wide variety of information about your market. You can then enhance your position by making the most of your strengths.

3. Know how to take a step back

A risk when you dive deep in conversion optimization, is to become blind to your website flaws. By working on it every day, it’s easy to become obsessed by numbers and not pay enough attention to the always evolving trends and needs. You get what is called the “curse of knowledge”, a cognitive bias.

To get a fresh outlook, ask relatives, new colleagues or hire users through companies like usertesting.com to navigate through your website. Create scenarios and watch them: do they easily find what they were looking for, what do they feel, is everything clear?

It’s capital to step in your users’ shoes, try to understand their thought process and how they engage with your website.

4. Analyze your conversion funnel

Go through your entire conversion funnel, step by step, from the Adwords Ads to your confirmation email. You can then identify where the failure points are.

From experience, we saw evidence-based recommendations produce much more significant results than those relying on good practices.

First, schematize your funnel by recording every action a prospect must take to become your customer. It will help you identify unnecessary steps, or missing items such as sponsorship program, CRM cycle, customer support, customer reviews, etc …

Then, open up your web analytics tool (or ask your analyst) to analyze and detect the steps with a high dropout rate.

Focus on those and similar areas on your website with the highest overall impact on your objectives, as well as those easily improved.

In our next (and last) article, we’ll finish our overview of A/B Testing methodology and best practices.