Conversion

Using A/B Testing to Optimize an Online Store

Larger online retailers have big budgets to optimize their navigation and user interface to drive higher conversion rates. They do this by experimenting relentlessly to find the designs with the results they are targeting: drawing visitors to their stores, delivering the products they are looking for, and ultimately closing the sales. They use analytics to help select everything from typefaces and call-to-action colors to page layouts and search result presentations.

The days of these types of decisions made by graphic designers are long gone. Today, extensive A/B and multivariate experiments using thousands of variables are performed by leading online retailers each day. Companies make decisions based on the results of the testing. If Page A delivers a higher conversion rate than Page B, use Page A regardless of one’s graphic design intuition. Increasing a purchase conversion rate by even one-tenth of one percent may dramatically improve profits. So companies test everything that could influence shopping behaviors.

Unfortunately, most small online retailers do not have the staff or resources to outsource the design of multiple versions of their web pages. The actual hosting and serving of multiple pages may be technically challenging as well, depending on their shopping carts. This puts them at a huge disadvantage to large retailers who can justify enormous investments in optimizing their stores.

This article examines some tools that you may want to consider to be more competitive.

Affordable Testing Tools

There are several tools for A/B testing that include “What You See Is What You Get” (WYSIWYG) editing as well as the hosting and analytics required to do sophisticated — or very basic — A/B testing. Most of the tools do not require any technical background. You simply redesign a page in their platform, select goals that will be used to measure the various versions, and insert some minimal JavaScript in the web pages you are evaluating.

Let’s start with Google Content Experiments. It is free and located in Google Analytics. It will allow you to test multiple pages against goals that you set up using a random sample of your site visitors. This is a good place to start if you currently do not test at all. It will require you to build multiple versions of your pages and host them within your shopping cart.

More sophisticated packages allow you to quickly build and deploy pages — without coding — outside of your shopping cart. Small business packages from Optimizely, Visual Website Optimizer, Unbounce, and Convert start at less than $20 per month with no long-term commitments.

Optimizely Example

Here is an example that I created using Optimizely. I spent less than five minutes creating an account and making the changes to the page that you see below. This is the original page, before changes.

Zoom Enlarge This Image

Original page before using Optimizely.

Original page before using Optimizely.

I switched the image of the girl to the right side, reduced the size of that section of the page so more of the page will appear above the fold. I also modified the Latin text in the second paragraph.

Here is the final result.

Zoom Enlarge This Image

Test page after using Optimizely.

Test page after using Optimizely.

The tools within Optimizely allowed me to simply drag and drop the image, resize it, and grab the entire text container and move it to the left. To edit the text, I used the WYSIWYG text editor mode, though I could also have used the HTML editor mode to add hyperlinks or other HTML markup. Here’s what that editing tool looks like.

Zoom Enlarge This Image

Optimizely editor.

Optimizely editor.

To test this page versus the original one, I could have activated that experiment within a few minutes. I could have set up several other versions of the page and included them in the same experiment. Other changes might have been to the fonts, the colors, or the image itself. I could have set a goal of a click-through rate from that landing page, to a product page, and eventually to a sale. Visitors to my store would be directed to one of the two or more pages in the experiment based on the rules that I established. I could have used geo-preferences, too, such as showing non-U.S. visitors a different page.

Optimizely hosts the experimental page and by adding JavaScript to my own website. Visitors would see my original home page or the experimental pages based on the rules I established.

In this case, the analytics are generated in Optimizely. Other platforms connect to third-party analytics tools, such as Google Analytics. Some also connect with marketing automation platforms like Salesforce and HubSpot for promotions, landing pages, and overall results.

Worth the Money

Optimizely, Convert, and Visual Website Optimizer all offer similar functionality and pricing. Convert positions itself for “experts” on its home page, not beginners like me.

As you upgrade plans, you can test more visitors, conduct more sophisticated targeting, and use multivariate testing and heat maps. Even mid-priced packages cost less than $100 per month. With some of the tools you will also find cross browser and responsive-design testing. Both are extremely important. Once you select the optimal pages, you will want to build and deploy them in your own shopping cart as these platforms are designed for short term experiments, not long term hosting.

Most of us are not A/B testing experts. In fact, I did no A/B testing when I owned my own ecommerce business. I wish I had. But my business was resource constrained and I was not aware of some of these tools. I would certainly try them today. It’s well worth a $20 monthly investment to try them. Many merchants will need the next higher level in pricing to receive the full benefits. But even at $100 to $300 per month, if you improve your conversion rates by as little as half a percent, it’s worth the money.

Dale Traxler
Dale Traxler
Bio   •   RSS Feed


x