If you've never heard of A/B testing, you're like many in media. Long the domain of ecommerce, A/B testing is the practice of testing two different variants to see which is most effective. 

"The slowest part of the A/B testing industry is publishing," said Dennis van der Heijden, founder and CEO of Convert.com, a testing software platform. They are last."

Getting readers to read more, share more, and click more are universal conversion goals for news and information sites. The practice of A/B testing can lead to dramatic increases in page views, social shares, and time spent on a page of online news and information sites, providing an edge over competitors.

I recently started testing my website, BargainBabe.com, which helps people save money. So I can vouch for the power of testing, while confirming that it took a lot of work and was not a panacea for my traffic stats. More technical knowledge was required for some tests that I tried to run, while others were quite simple.

So what's the catch? Testing can drain in-house IT resources or cost hundreds of dollars a month if you pay for a testing service. One expert said that even veteran testers may only see one out of three tests produce a winner.

Thinking you can increase your traffic stats simply by testing headlines and styling buttons? Let’s dive into what to test, what people mess up, and how to get started. 

What to Test

Headlines: If clickiness is your goal, here’s one place to start. Test two or three different versions of a headline and switch to the winner after sufficient traffic has determined onn. This is a relatively easy test to set up. Testing software provides conversation stats so you can be confident in your winning variations.

Share buttons: Social platforms drive a huge amount of traffic so it’s important that your social share buttons are as appealing and attractive as possible. Big or little? Bold or italic? Round or square? At the top or bottom of a post?

Links to more articles: The more time a reader spends on your site, the more ads are served. What is the most effective way to get readers to stick around? Links to more stories in the middle of a page? At the bottom? Linked images? 

Font size: Many sites are finding increased engagement with larger font sizes.

Above the fold: What kind of content needs to be above the fold to get readers to scroll down? Does one image lead to more scrolling than other? How do graphics or bullet points engage?

Comments: What styling and positioning leads to more comments?

Ads: What positioning and styling leads to more clicks?

Email subscribe box: What leads to the most sign ups? I found increasing the size of my email signup box and removing nearby distractions led to greater conversions.

Call to action: What is the call to action on each page? Is it a button to share the story? A box to join your email list? Links to more stories? Define what action you want the reader to take, then test the heck out of it. 

What People Mess up with A/B Testing

They trust their gut. You know your readers, right? Run a few A/B tests and you’ll realize that your gut is often wrong. Don’t believe me? Test your gut each week at WhichTestWon, which publishes successful A/B tests. "I can tell you what I think will look nice, you know what you think looks nice, and your [boss] will tell you what they think looks nice," said Andrea Warner of WhichTestWon. "All that matters is what your readers think. And the only way to tell that is testing."

 They expect every test to be a winner. A winning test is statistically superior to other variations but no every test has a clear winner." We did a study and we've seen that on average, people who are not a testing agency and that don’t have a team for testing, they only have 1 out of 7 successful tests," said van der Heijden of Convert.com. "The better you get, the more you do it, the more you learn from it. You can get it up to one out of three."

 They don’t test for long enough. Let each test run a week so your confidence in a winning variation is at the top of the ninetieth percentile.

 They copy other tests. "Every business and its users are unique, so what works for others may not work for you," said Paras Chopra, founder of VWO.com. "So A/B testing becomes successful only if you test ideas that you generate out of your own research, rather than randomly copying and testing other people's ideas."

How to Get Started

• Start with a hypothesis. Don’t test randomly.
• Get your team on board and check if tech help is available. I found a huge amount of tutorials and videos online. Also, a basic HTML/CSS class at CodeAcademy.com was helpful with terminology and structure.
• Consider hiring a testing agency or buying A/B testing software, such as Visual Website Optimizer or Optimizely. Both offer 30-day free trials.
• Keep testing. What worked a year ago may not work now.