A/B Tests You Should Run to Increase Sales


There are two ways to increase sales through internet marketing. Businesses can focus on generating more traffic to their sites, relying on the “law of averages” that says a certain percentage of them will make a purchase. Or, they can work to improve their site’s effectiveness. This helps them get more sales from the current amount of traffic.

Conversion Rate Optimization (CRO) focuses on the second method.

The whole point behind testing of is to assess the effectiveness of every part of a website’s sales funnel, which is essentially every page of the website so that a higher percentage of visitors make the conversions you want at each stage. Once you decide what to test, you keep the old version of the element, create a new one based upon a hypothesis you wish to test, and then evaluate which one converts more.

The concept is simple. Implementation? Not so much.

Featured Download: Want more advanced tips on how to optimize your landing page? Download our full checklist: 33 Point Perfect Landing Page Design Checklist right now.


The Big and the Little

Sometimes, just a small change can improve conversion rates – changing a word or two on a CTA button or just changing the color of the CTA button itself. One company changed the wording on its button from “Get Your Free Trial” to “Get 60 Days Free – It’s On Us” and realized an increase in conversions pretty quickly.

Big changes include such things as changing the layout and design of an entire page. In the middle would be changes such as shortening an opt-in form.

Whether you decide to make big or little changes in your testing, the point is this: you have to go about testing in a reasoned, systematic way, just as a researcher in a chemistry lab does.


The Value of Testing

Value of testing

Suppose you own a business that sells an HR software package for $200. You are not getting the revenue that you want, and you have the decision to make. You either advertise more, or you evaluate your site and see what might be changed to better convert existing traffic.

Right now, as you analyze your ad budget, you determine that it is costing you about $10.00 for each customer you are getting. So, if you have 50 customers, it has cost you $500 to get them. Now, you can choose to spend more on advertising and hope that you will get enough of an increase to keep your cost at $10.00 per new customer, although there is no guarantee of this.

Or, you can take a scientific approach to your site-based conversion efforts and see if some changes might achieve the increased conversion goals. The value of doing this is as follows:

  1. If you get a significant increase in conversion rates, then you can lay off of the additional advertising costs.
  2. The increase in conversion rates is probably going to be more permanent. The testing may have cost you the additional $500 that you might have put into advertising, but it will not have a “shelf life.”

So, let’s say that you choose the A/B testing route and you end up with 75 conversions as opposed to the 50 that you were regularly getting each month. Now your acquisition cost of each customer had gone down to less than $7.00. And your monthly revenue has gone up from $10,000 to $15,000.


Deciding What to Test


This is the big question, of course. If you rely on your “gut instinct” in making decisions about split testing, you will do a lot of ineffective testing that will yield no measurable results.

Another mistake you may make, in attempting to save money, is to copy the testing that other sites have done and found successful. There are case studies of testing all over the web, and you can certainly check them out, using their results to make changes without testing – big mistake. Their products/services are not the same; their audiences are not precisely the same. Using their testing results is not a valid testing outcome for your business. Just because one business changed the color of a CTA button, it does not inform what you should do. The color may be a good match for their theme and design, but not for yours. You have to test yourself, or you will never know for sure what will work for you and your audience.

As you make decisions about what to test, you need to establish a step-by-step process.

  1. What are your conversion goals? Are you wanting an opt-in for a newsletter subscription or some other conversion that will get you an email address? Do you want visitors to take advantage of a free trial? Or are you driving traffic to just a purchase without intervening conversions? Your decisions about testing should focus on the elements you have on the site to get those conversions.
  1. Heat map technology is a great tool to begin making your decisions about what to test. With this type of analysis, you can determine where your traffic is bouncing rather than following through on a conversion. It will allow you to target certain pages or places on certain pages that need to be evaluated and tested.
  2. As you look at the bounce spots on your site, think about what might be changed. Is there a conversion button? Is there lengthy text that could be shortened and replaced by a visual? Is there a lengthy opt-in form that can be revised and shortened, asking for less information from a visitor? There’s a reason for those bounces, and you want to identify it and make changes.
  3. Develop your hypothesis around one element only. Multivariate testing is far more complicated and you would need a statistician to crunch the data. Let’ say you have decided, based upon bounce rates, that you are going to change the opt-in form for a free trial. This is the only thing you will be changing. So, you reduce the number of fields and maybe make the entire form larger, so that mobile traffic will have an easier time. You run your test for 3 months and then evaluate the results.


Evaluating those Results

Evaluating results

Now, suppose the 3 month period of testing has expired and you have seen a 19% increase in opt-in for sign-ups. That is actually a pretty good percentage. But you are not sure whether the shortening of the form or the enlargement of the fields is the reason. Does it really matter? If your goal was a 20% increase in opt-in conversions, you have just about met it.

You could dig deeper and then do A/B testing between those two elements you changed, but then you can end up testing the smaller things to death, and the results may not be significant. You need to focus on the goal of your testing, rather than just the micro-conversion changes you might get by drilling down further. Better to move on and test the next element on your list.


Always Test and Measure with Your Final Goal in Mind

Final Goal

The final goal for any business is increased sales. So, let’s go back to that opt-in form for a minute. You have determined that one version gets you a 19% increase in opt-ins. Now, you need to understand how many of those opt-ins eventually became paying customers after taking that free trial. If you are not getting the conversion level you want, then you know what testing must come next. What stage of your sales funnel addresses this? It’s time to analyze it and set up additional testing for that stage.


Test Only those Elements Directly Related to Conversions


There’s a point of view out there that says every time you change the smallest thing on your website, you should conduct A/B testing. This is a myth. If you change some images or some text on your site, you may want to test to make sure that those changes have not impacted speed, and that is legitimate testing. It is not, however, A/B testing, and that is not necessary.

When you change copy, visuals, or elements on a page that are supposed to result in conversions, however, that is directly related and should always be tested.

Let’s take your blog as an example. Most people do not think of a blog for A/B testing. Its purpose is to bring in traffic, not to convert people, right? Wrong. Good blogs have conversion elements – signing up for email delivery of posts, linking to places on your site to get value-added elements, or just social sharing, if that is one of your longer-term conversion strategies. You may have engaging content writing, compelling headlines, and lots of traffic, but if those desired conversions are not happening, then something is wrong. This is an area you will want to evaluate and see how you can change those conversion tactics. Maybe placing conversion opt-ins in the middle or throughout the posts will make a difference. Maybe changing the CTA’s on the rails or adding testimonials will improve conversions. Your blog is certainly related to conversions, so do not ignore it.


So What A/B Tests Should You Be Running Right Now?

A/B Testing concept

No one can give you a one-size-fits-all answer. But here are the factors that will help you make these decisions.

  1. Identify your conversion goals for each stage of your sales funnel.
  2. Evaluate and A/B test only those elements which directly relate to your conversion goals – everything else is irrelevant
  3. Use technology to identify which of the conversion elements are not performing well.
  4. Set up testing very carefully and systematically

You don’t want to waste time and money testing things that will not really result in increased sales. If you follow these suggestions, you will get the right data.

Do you have any other A/B testing strategies that you would recommend? Let us know in the comments!

Filed Under: A/B Testing

About Norman Arvidsson

Norman Arvidsson is an independent author who was born in Sweden but then moved to the United States with his family. Now his goal is to share his experience with others through blogging. You can contact him through his social Twitter (@NormanArvidsson).

Back to posts
Read previous post:
close up of creative team
Do You Have a Sales Content Problem? Here are 8 Ways to Fix It!

Many equate sales content to content marketing. But sales content is actually a distinct and essential part of your content marketing strategy.   Sales content...