[vc_row][vc_column][ultimate_heading main_heading=”A/B Testing” heading_tag=”h1″ spacer=”line_only” spacer_position=”bottom” line_height=”1″]Making Educated Decisions Based on User Interaction Data[/ultimate_heading][vc_empty_space][/vc_column][/vc_row][vc_row][vc_column][vc_column_text]
A website is never finished.
Building a website is kind of like buying a new house. There will always be something you or your Orlando web design company can improve. The projects never end.
A/B testing, which is also called split testing, is a great way to collect data before making a permanent change you think would be beneficial to your website. This method of testing allows you to compare versions of your website by running two or more variations at the same time. The experiment, if run correctly, will show you which variation performs better for a predefined goal.
How it works.
Version A is usually the control in the experiment. Version A will represent the current state of the element in question. For example, if you are testing the conversion rate of a contact form based on the color of the submission button, Version A would be the current color of the button. Version B would represent the proposed change. The button is red on Version A and Green on Version B.
Using some kind of A/B testing tool or plugin you would run the two simultaneously while collecting the data from each based on the number of conversions in a set period of time. Each would be shown to an equal number of website visitors.
Before running the test you should know exactly how long you will conduct the experiment and what you are tracking.
Whatever tool you choose to implement the test should have some kind of analytical dashboard to collect and measure the data. The one with the better conversion rate at the end of the experiment is what you will make permanent on the site.
Change it up! But not too much.
Sometimes the simplest things can improve conversion optimization on a website. Let’s use a contact form as an example. Your form may have a heading that says “Contact Us” above it. Maybe if you changed the wording to something more specific you would get more leads? So, using the same form, you would change the heading to “Contact an Industry Specialist”. Run the two side by side using A/B testing and see which produces better results. You may find that the original version was best. In that case you can either stick with it or try another variation.
It is best to only change small things, one at a time to do this testing. If you try to change an entire page and run two completely different layouts you will have no way to know what part of the change was better for conversions.
For example, try swapping out things like:
[/vc_column_text][dt_vc_list dividers=”false”]
- Images
- Headings
- Calls to action
- Button colors
- Supporting content
- Testimonials
- Social proof
- Awards and badges
- Media or press
[/dt_vc_list][vc_column_text]You May Also Like: Website Speed Optimization[/vc_column_text][vc_column_text]
What’s in it for me?
A/B testing is the best way to make the most out of your existing website traffic. The cost of increasing the traffic to your website with methods like SEO or PPC can cost a pretty penny. A/B testing helps you get the most out of that investment, ensuring you’re site is a converting machine. This is a method we often implement with our SEO clients.
A/B Testing do’s and don’ts for SEO.[/vc_column_text][dt_vc_list]
- Keep your cloak and dagger at home. Cloaking is a when you show humans and googlebots different versions of the same thing. This is frowned upon and clearly against Webmaster Guidelines. Going against those guidelines can get you removed from Google search results.
- 302s are cool, 301s are not. When we’re talking about A/B testing and you want to redirect the main URL to the temporary variation of the same page, 302 redirects are the way to go. 302 is a temporary redirect, 301 is permanent. 302 designations tell search engines that the redirect will only be in place for a short period of time so that the URL will not be indexed. The redirect should only be in place for the time of the experiment.
- Don’t run the experiment for too long. The perfect amount of time is going to vary based on the amount of traffic the site receives and your conversion rates. Once you feel like you’ve collected enough sample data, remove the temporary URL, scripts or anything else you’ve used for the testing. Put it back to the way it was originally while you analyze your data.
[/dt_vc_list][vc_column_text]
Be patient.
You’re going to want to make changes as soon as you get your data back. Take a breath. You’re not finished yet. From start to finish your A/B testing procedure should look something like this:[/vc_column_text][dt_vc_list]
- Analyze your current website data to find weaknesses. Where is your conversion funnel lacking? You can use tools like Google Analytics and Mouseflow to find this information. For example, if you’ve had the same call to action on your homepage for awhile you might want to see how many people actually act on that. Mouseflow, which provides recording of visitor interactions, can also show you if people are moving through the website as you intended.
- Come up with a hypothesis. For example, we believe our call to action is not drawing the attention of the user. That call to action is in slider image on our homepage but people just aren’t clicking on it. We believe that redesigning our homepage slider image will increase the number of people that click on it, sending them into the sales funnel.
- Time to test. Create the new homepage slider image and A/B test the original page against the page with the new slider image. Collect enough data to see a clear difference in the two.
- Analyze your data and come to a conclusion. The one with the better numbers wins! You may now implement if there is a clear winner. If not, go back and rework your hypothesis and try again.
[/dt_vc_list][vc_column_text]The most important focus with A/B testing is your goal. What are you trying to improve? Will the change you’re proposing accomplish that goal? Testing variations is the best way to find out.[/vc_column_text][/vc_column][/vc_row]