A/B Testing ideas
A/B Testing, Conversion Rate Optimization

A/B testing guide

What is A/B Testing?

A/B testing is a method that allows e-marketers to test their ideas before asking the programming guy to make any changes in the website’s code. An A/B test runs like this: you’re comparing the original version of your website, A, against the new versions that you get after making a MAJOR change, B and C, let’s say. It doesn’t matter how many versions you get; A/B testing is called like this because it splits the website traffic. The visitors become receivers of different versions of a webpage. You can run the test on all your traffic or on just particular segments of traffic that interest you the most.

A/B Testing ideas

Either the element you’re changing is an image, a different size of the call-to-action button or a longer form on a landing page, it has to be a MAJOR difference so the A/B test can lead to relevant results. For more particular goals like analyzing the visitors’ behavior on a website, you can use another type of test, the Multivariate test, that allows you to make multiple changes on a webpage, not just ONE.


Before you decide what major change to test, start from a hypothesis. A/B testing is a scientific method, so it gives best results when it is correctly implemented. There are two approaches when developing a hypothesis for an A/B test:

  • the data approach that uses statistical data(Google Analytics is a great tool)
  • the people approach

For example, if you notice a high time on page and a high bounce rate* it may be a problem with an incentive on the webpage. Probably the call-to-action button doesn’t work so well. Now that you’ve looked into data, you know that the call-to-action button is the element that needs to be changed and tested.

*time on page=the average time spend on one particular page*bounce rate=the percentage of visitors that leave your website rather than continue to visit other pages of it

What to test?

Headline: is the first impression a visitor gets when landing on a webpage. Depending on the webpage goal, you can test: positive vs. negative, long vs. short, single headline vs. primary and supporting copy or sub-headline and so on.

Images: have a huge impact on the majority of people. That’s why you should use high-quality images that drive the visitors’ attention to a call-to-action. Like the headline, images offer an idea about what the offer is about(in case of a landing page for example). What are the major 2 variations that can be tested?

  • Masculine vs. feminine
  • Color vs. black & white
  • Product vs. people
  • One image vs. multiple images

CopyShort and concise pages work better. People spend from 10 to 20 seconds on a page, according to Nielsen Norman Group, so they won’t bother to read long paragraphs. Either you choose a long or a short text, make sure you organize it into clear paragraphs that are built starting from a proposition that reveals a benefit or a feature.

Colors: It may be about how colors influence your visitors(how men perceive colors is different from how women do), but it may also be about the contrast. You don’t know until you test.

Call-to-action: The role of a call-to-action is to get conversions. If it is not working it means that you should change it, because this is the key of your conversions. Actually, it is your conversion.  Make sure your call-to-action reinforce the benefit introduced in the headline.

Social Proof: From all the forms that social proof can take, testimonials and endorsements are the most suitable for an A/B test. You can choose between one standard approach for all the visitors or more testimonials for different segments of traffic (e.g.: for women a testimonial with a smiling face of a man, for men a testimonial with a smiling face of a woman)

Myths about A/B Testing

A/B testing best practices

1.A/B testing takes a lot of time

So not true! It takes only 3 minutes to implement a javascript code and another 20 minutes to deploy some ready-to-use templates based on your strategic hypotheses. That’s how Omniconvert works. You can also create a survey in order to find out from your real visitors why aren’t they converting as much as you predicted to.

2.I don’t know what to test

You can test anything. Take your most important pages that are the most vital for your business; it can be:

  • a landing page where you want to get micro-conversions: gathering data fom your visitors by filling in a form so they can download a free whitepaper, guide, e-book and so on
  • the shopping cart page: in case of e-commerce stores, this is the Achile’s Heel: if you don’t catch your potential customer here, he will leave and may never return back to complete the purchase
  • anywhere on your website where you feel to

3. I need programming skills

Advanced A/B testing platform like Omniconvert use a WSYIWYG Editing tool, so you don’t need a programming background to change the placement of a call-to-action button, an image or anything you want to change in your website’s design.

4. I will screw things up!

Not at all! You can test changes on a small amount of traffic. Choose your audience based on your goals. It can be only 5% of your traffic and the test will show you which version led to the highest conversion rate.

5. It takes a lot of money

This is an unfair assumption. You can use affordable platforms like Omniconvert, create an account by filling in your name and email..and that’s it. You can optimize your conversion rate based on actions tested on real visitors in real time.


Leave a Reply

Your email address will not be published. Required fields are marked *