Update your browser, this website may not display correctly.

Increase your Website Sales Performance with A/B Testing

Converting your visitors is what your website is all about, so how well do you do it?

Can you predict that for every $1 spent on marketing through Facebook, Instagram, or any other channel, that you will earn X cents? If so, are you happy with the value of X, or do you think it can be larger? If not, do you know how to build such an equation?

As you begin 2018, here are 5 steps to increase your website’s sales and make better decisions through A|B testing.

1. Measure with Analytics

Making the right choices means understanding details and how they fit into the big picture. Try to avoid Buzzwords and vanity metrics and define what is the right set of indicators to pay attention to – there are tons of fantastic tools out there. You only need to start with one to measure Key Performance Indicators (KPI’s) for your specific business.

Google Analytics (GA) is the oldest known platform out there, while HEAP and other more modern tools exist as well – it all depends on what you need to measure and keep track of. Now, I love a good old fashioned set of data in GA – it gives you the basics – how many times your site was viewed, what your sales were, and through it’s Behavior Flow feature, you can build a map of the customer journey.

One of my favorite clients simplified their customer journey by drawing boxes for each step of their funnel and marked it as such – smiley green faces for the pages with the desired conversion rate and red sad faces for the pages that converted less than desired. This made prioritization smooth – our testing focus should result in removing sad faces. It also communicated our program easily with decision makers and the team around us.

2. Test Your Copy

Do you have doubts about the content you write? Perhaps the thought of “what if I had said [this], instead of [that]?” crosses your mind after making changes to your website, publishing an article, or sending an email?

Headline testing is used by CNN, the Weather Channel, The New York Times, and more. Shouldn’t you? Copy tests are some of the lowest hanging fruit you can go after.

One of my favorite clients simplified their customer journey by drawing boxes for each step of their funnel and marked it as such – smiley green faces for the pages with the desired conversion rate and red sad faces for the pages that converted less than desired

3. Test Your Customer’s Emotions

If you’ve ever worked sales person to person, you know that selling is based on how your customer feels – do they believe in your product? It’s an emotion, not a technical spec. There are aspects of your website that you have complete control over to stimulate your customer’s emotions and make the sale – from colors to imagery, funnel order and more.

When we look at the product being sold, how does it make someone’s life better? How does your customer want to feel when using your product?

4. Feature Tests

You can use a great, sophisticated product like Optimizely to save yourself thousands of dollars and hours from building useless features.

For another one of my clients, we routinely run small to medium sized feature tests built quickly to see if it’s worth implementing and testing further. Larger feature tests are tested differently which I will discuss at another time.

Here’s a very quick, small feature we tested – the Head of Design proposed desktop users would navigate with the keyboard to scroll through a newsfeed we created around the impact the company was making (left former article, right next article, as it scrolled from right to left). The CTO had his doubts and thought it was a very low priority for his team to develop – he asked me my thoughts, so I said let’s test it!

What do you think happened? Did users use the left and right arrows intrinsically?

I ran a test that had Javascript listen to the press of the left and right arrow keys for desktop users and would change the story shown when pressed. Simple enough and the test was up that day.

After filtering the company’s IP address (and my remote office IP address) from the results of the test, there were less than a handful of key presses. The feature worked but was used by so few that it was not necessary to implement full time. We left the test running for a few more weeks so the Head of Design could have the feature. I checked back, and even the usage by people in the office dropped to null.

Months later, when we rebuilt the feature and went to a second version, I was consulted on what usability features to include. The data showed not to include keyboard navigation.

5. Talk to Your Customers

Before you do any of this, make sure you’re open to feedback. Listening is your most important tool to understanding what problems exist and then to solve the right thing.

Simple survey tools are all over the place and whether you’re using usertesting.com or a google form for a quiz, here’s four simple things to include:

  1. Add context to a user’s feedback. Ask who they are and identify them by segments important to your business (anything from age, location, gender, to how much they spend on your products and how often)
  2. Don’t ask leading questions (i.e. avoid ‘do you LIKE A better than B’).
  3. Always leave room for 1 open ended response for anything else the customer would like to say. Sometimes, this provides some real pearls of wisdom. Other times, a great laugh.
  4. Don’t forget to throw in a Net Promoter Score (NPS) question at the end! How much each user loves you is the best context for their feedback.

If you want to learn more on how I run Optimization programs for clients and how you can run one too, come to my presentation on “Increase your website conversion: A/B Testing 101” on March 8th at 6:30PM in San Francisco, CA. See you then!

Gary Marx

Gary Marx is a Full Stack Engineer, running website optimization tests for many companies and project management for the testing process and strategy.