Improve Sales With A/B Testing

Every business website has a “conversion goal” — a desired end result for each visitor. E-commerce sites exist to convert visitors into customers. News sites exist to convert visitors into subscribers. Other sites exist to convert visitors into trial users and eventually paid subscribers. A/B testing is a critical tool to get more of those desired results. Here’s how it works…

Test, Test, Test!

Converting a visitor into something else is the reason a business website exists, the reason you pour so much effort and money into it. So it’s critical to measure how well a site does its job and to constantly improve its performance by testing variations on the design.

Direct mail marketers have done this for decades. They’ll send one batch of mailings that says, for example, “Open Now for Big Savings” on the envelope, and another batch that says “Your 20% Off Coupon is Inside…” The one that performs best wins, and a new round of testing begins. Once they fine-tune the envelope, they’ll move on to testing the elements of the offer inside.

If you’re mailing to thousands or millions, a simple tweak of a headline, an underlined or bolded phrase, or the choice of one product photo over another might turn out to make a huge difference in sales. And of course, the same concepts apply to your website.

A/B Testing Tools

The overall conversion rate of the entire site is noteworthy but it doesn’t tell you what, exactly, can be improved. If you have relatively few and unique pages you can measure the performance of every page that asks a visitor to take another step along the path to conversion to the desired status. Landing pages, product presentation templates, shopping cart checkout pages, and so on can have their conversion rates measured and improved.

But if you run a sales site with thousands of products, measuring the conversion rate of every single product’s image and text description is probably not cost-effective; fortunately, you don’t have to do that tedious job.

What Should You Test?

A few key pages in your site are probably more critical than most others. Landing pages, the first ones that visitors see when they enter your site, are critical. Small improvements to a landing page’s conversion rate can make highly profitable differences in the site’s overall conversion rate. You may want to review my article, How to Make Great Landing Pages and then experiment with your site’s landing pages. Other critical pages include “buy/try it now” pages, and pages where you ask the visitor for payment information.

The headlines and sub-heads (your H1 and H2 tags) on a sales page can make or break you. Testing variations of the wording, color and size of headlines may yield surprising results. The same applies to your Buy Now button. Should it be red or orange? Square or oval?

A simple way to test changes to a web page is called “A/B testing.” Two versions of the same page are shown to visitors for a certain timeframe, and the one with the higher conversion rate (the most sales, clicks, etc.) wins. A/B testing is simple in theory, but how do you do it in real life?

There are online services that handle the heavy lifting of A/B testing. You just supply the URLs of pages A and B, and the service tracks where visitors to each page come from and what they do after landing on pages A and B. You get a report that includes the two pages’ conversion rates. Ta-da!

Automated A/B Testing Tools

Visual Website Optimizer is a marvelously simple, straightforward A/B testing service. It includes a well-written guide to the entire process of A/B testing. Two free trial formats are offered, and both include all the features of the paid version. One requires no credit card info but is limited to 1,000 visitors. The other is a 30-day trial providing stats on an unlimited number of visitors, but credit card info is required up front and you’ll be converted to “subscriber” status unless you cancel within 30 days. Annual fees range from $9/month for up to 2,000 visitors to $499/month for an unlimited number of visitors.

Optimizely offers a permanently free service with limited metrics; for example, it includes a visitor’s referral source (the site from which he came) and browser type, while the paid version adds the ability to track by marketing campaign (the visitor not only came from but from Ad X on, the device that a visitor is using, and even custom segments. Prices are quoted on a case by case basis. Both free and paid versions include automated site optimization and mobile versions of one’s site.

If you use Google’s Adwords or Adsense tools, there are testing and analytics tools built into those platforms. Use them to fine tune your ad copy, placement of ad slots, and other elements of your pages or campaigns.

There are many other A/B testing services out there. VWO and Optimizely, to me, represent the extremes of transparent simplicity and opaque complexity, respectively. I’m always wary of a company that can’t tell me what its product costs until it knows everything about my business. But Optimizely’s list of advanced features will give you an idea of how sophisticated A/B testing can get, if you have the time and money.

Oh, and don’t forget to A/B test your emails. The Subject line is the most important factor when it comes to persuading people to open your email. Maximizing the number of people who then subsequently click a link to your site is equally important. Email marketing tools like Aweber and Constant Contact provide tools to help you measure open rates, click rates and other analytics.

Testing your designs to maximize opens, clicks, sales and other desired responses should be an ongoing task for any serious online business. If you keep at it, each improvement will build on the last, resulting in a bottom line that’s continuously rising.

Your thoughts on this topic are welcome. Post your comment or question below…

More Posts about Ecommerce:

  • Yes, You Need an Ecommerce Platform

  • New Square Reader Does Apple And Chips

  • Are You Paying For Ads That Nobody Sees?

  • Improve Sales With A/B Testing

  • Are Trust Logos Worth It?

More Posts about Webmaster:

  • 400 Free Tools For Business Building

  • Is Your WordPress Site Vulnerable?

  • Improve Sales With A/B Testing

  • Google Webmaster Tools

  • Adding Search To Your Site


  1. Mac and Cheese says

    Hi, Bob,

    I ready your essay on A/B testing with interest. There is probably a minimum number of test subjects, but I don’t know what it is. For example, if I’m mailing to 10 prospects, I wouldn’t think it would make sense to do a split test. The results probably wouldn’t be repeatable. But a list of 10,000 prospects? Maybe so.

    What, in your opinion is the minimum number of tests? 1,000? 5,000?


  2. Bob Miller says

    This article was really usefull, AB test can improve sales.
    But most A/B tester is not easy to understand to whom are not confortable to interpreting results.
    A tool I suggest is :
    Because people should really understand the results coming from an AB test.

Join the Conversation! Leave Your Comment...


Free Small Business Tech Support -- The Rankin File
Subscribe to The Rankin File: Free Newsletter
Copyright © 2005 - Bob Rankin - All Rights Reserved
Privacy Policy -- See my profile on Google.