If your small business uses online sales letters or squeeze pages, you're probably thinking about how to increase sales conversions. The wording of these pages is important, since you obviously don't want to say anything that will turn away potential customers. But how do you know what to say?The most popular methods of testing are A\/B testing and multivariate testing. With A\/B testing, you test two versions of a Web page to see which one produces better conversions. The "winner" can then be used as a control page against which to test a new variation. Multivariate testing, on the other hand, tests multiple pages and multiple combinations of page elements at the same time to see which one converts better. The Google Content Experiments (which replaces Google Website Optimizer), for example, lets users test up to five full versions of a given page, each administered by a different URL.This article will specifically look at A\/B testing, since it only involves a single variable. Plus, you can think of multivariate testing as a set of A\/B tests being performed at once.A\/B Testing Variables to ConsiderWhen looking at a sales page or squeeze page, there are many things you can test.First and foremost is the headline. According to Magnetic Headlines, 80 percent of visitors will read a headline, but only 20 percent will read the rest of a page. The headline is the first thing that potential customers see, and it's the one page element that can make or break sales and\/or opt-ins.The call to action matters as well. Many marketers test this with different button types, text and colors to see if this will improve conversions. For reference, check out The Belcher Button.It's also important to test different pricing models. For example, customers might balk at an up-front price of $497, but they could be happy with a monthly fee of $47, even though that means paying more over the course of a year.Finally, consider a video sales letter and test whether it outperforms a traditional text-based sales letter.How to Write Sales Page, Squeeze Page CopyBefore you begin an A\/B testing strategy, you need to write the copy that will go on the pages you're testing. Brandon Zundel has some advice for getting through the writing process."Whenever I feel like I'm working toward a final draft, the next thing I do is read it out loud. When you read it out loud, you'll find yourself stumbling over words and phrases," he says. "If you stumble while reading your copy, what do you think will be happening to your readers?"Great copy should flow like a steep stream. When you can breeze through it without any hiccups, that's when you know you have it worded right," Zundel continues. "Once I've gotten it to that point, I'll let it rest for a day or two, then I'll repeat the process again."How-to: 7 Email Marketing Tips to Gain Customers in 2013It's a good idea to give marketing colleagues access to your materials (assuming they are copywriters). Often, they will spot problem areas that you've missed and can help you determine what needs to be changed. You can also survey your clients (assuming in this case that you have an email list) and ask for their feedback. You may be surprised by what they tell you.If you want to learn more about copywriting, I recommend the resources of Internet marketers Clayton Makepeace and John Carleton, an American Writers & Artists Inc. (AWIA) online copywriting course or the newest edition of the book Web Copy That Sells, which is due to be released in February 2013.Incorporate Targeted Traffic Keyword Research Into CopyYou may not realize it, but another important consideration when A\/B testing is traffic\u2014specifically, targeted traffic that matches your product or service. Knowing what keywords or phrases bring visitors to your website, and using those words on the pages that are part of your A\/B testing strategy, will improve your website's search engine optimization and make it visible to more people.You can obtain traffic from a variety of sources, both free and paid. Free traffic can come from articles, videos, LinkedIn forums, Facebook, Twitter, referrals and so on. Paid traffic can come from Facebook ads, Google Adwords and placement in online or print publications that match your target audience. If you're considering paid traffic, I recommend the work of Perry Marshall, who has written about both Google Adwords and Facebook ads and whose website offers some free content as well as recommended software tools.How-to: 20 SEO Tools to Help You Achieve, Measure SuccessTo get that traffic, you need to do keyword research. One common (and free) option is the Google Keyword Tool, which is part of Google Adwords. Another option is a paid solution. My keyword research tool of choice is Market Samurai, which I find an excellent resource for keyword research and testing. It's an excellent resource.Once A\/B Testing Begins, Aim for Conversions, Not Page ViewsWhen it's time to test a sales letter, test version "A" and "B" at the same time, not separately. This means one visitor will see sales letter "A" and the next visitor will see sales letter "B." The A\/B testing software you use will vary the sales letters for you; software options include Maxymiser, Optimizely and MailChimp.When testing, you need to let a good amount of time pass in order for the testing to produce results. Remember that you're looking for conversions, not page views. Marketing opinions vary, though one useful benchmark suggests waiting for a minimum of 100 conversions before making conclusive judgments about your copy. This isn't a hard and fast rule, though, as it might take many more conversions\u2014possibly 1,000 or more\u2014before you'll see a clear pattern. This could take a week or more, depending on how much traffic your site typically gets.Remember, too, that each change you make to your copy will increase the need for testing. Going back to the benchmark of 100 conversions per change, making three changes to your page would mean a benchmark of at least 300 conversions. This becomes even more complicated if you use multivariate testing: If you have five pages and you create five variations of each, then you're looking at a minimum of 2,500 conversions. Sometimes, the final results are conclusive. When Zundel ran an A\/B test on the phrases "get my money" vs. "get my cash," for example, the former was the clear winner by a 30 percent margin. If, on the other hand, it's too close to call, tweak the "A" and "B" versions and try again. Whatever the case, ignore your gut feeling and go with what the statistical data tells you.For further reading, ion interactive and Smashing Magazine provide additional insight into the A\/B testing process.After A\/B Testing, Work to Further Improve ConversionsYour A\/B testing strategy shouldn't end just because the A\/B test itself has concluded. Use the results, as well as your previous keyword research, to further improve your sales page and squeeze pageAdd a clearly worded guarantee and return policy to your copy. (In fact, certain credit card processing services, including ClickBank, require this.) It's a good idea to include a graphic as well. \nUse case studies and testimonials\u2014these show potential customers why your current customers appreciate your products and services. Include the full name, address and location of the client, both for verification and so prospects can speak to your customers themselves. Video and audio testimonials, which you can place within your case study text, can be a nice addition.\nWhen you post your contact information, use a real phone number that people can call and add a photograph as a opposed to an icon. I've had great success with this approach.\nMake it as easy as possible for a potential customer to buy your product or service. Shopping cart abandonment is a common problem for online businesses. The reasons are many, though one major cause is a complicated ordering process.\nOffer a short, free consultation on any aspect of your service. While you might fear an inundation of buyers taking advantage of this, my experience tells me that few people will actually do it. \nFinally, reduce a customer's fears of buying online. My marketing mentor, Jeff Mulligan, created a video about that subject and added a link to it in his copy. By adding that personal touch, his conversions went up.As you'll discover, the best way to learn what works\u2014and what doesn't\u2014is by testing many different aspects of your copy, whether it's on a sales letter or squeeze page. When one variation is a clear winner, use that as the control version and test a new variation against it. The more you test, the better your pages will perform, and the more conversions you will ultimately get.Nathan Segal has been working as a freelance writer, photographer and artist for 14 years. He is based in British Columbia, Canada. Reach him via email or visit his website. Follow everything from CIO.com on Twitter @CIOonline, on Facebook, and on Google +.