What Is A/B Testing and How Can It Help You?
Marketing and running a business are very dependent on human behavior.
As business owners, we want to know what attracts someone to our business and what will make them choose us over another business.
If we run a physical store or restaurant we can see in person how different decorations or coupons influences customers' behavior. But if our company exists only on the Internet, it can be much harder to understand what engages people.
There is a way to test this and it is called A/B testing.
“A/B testing produces concrete evidence of what actually works in your marketing. Continuously testing your hypotheses will not only yield good results for conversion rates, but will also give you a better understanding of your customers. Having a clear idea of what your customers actually like and prefer can do wonders for your branding and marketing in other channels as well.”
--Jaan M. Sonberg, How to Build a strong A/B Testing Plan That Gets Results, Conversion XL
It works like this:
- You change something on your website.
- You direct some of the visitors to the new version and some to the old version to see which version has the highest conversion rate.
- You keep whichever version has the highest conversion rate.
Sounds easy, right?
It can be easy, and it can be highly effective -- when you do it right.
But as with anything, A/B Testing can become a bit more complicated when you try to do it yourself.
So how do you actually do it?
A/B Testing is in many ways like conducting a science experiment.
You create a hypothesis based on existing data, such as, "Changing one or several elements of my landing page design will optimize its conversion rate."
From there, you begin a test in which you compare two variables. Examine the data to see which has the best outcome.
Conclude whether your hypothesis was correct or incorrect. Plan future tests around this conclusion.
The difficulty here is that when you do A/B testing, you are looking at human behavior. Sort of how kids learn to behave in school to get the teacher to like them, you want to figure out what influences your customers' behavior.
There are five steps involved in determining which page design will make your web visitors "like" you the best.
- Analyze Existing Data
- Plan Changes for Testing
- Make the Changes
- Start the Testing
- Analyze Your Results
1. Analyze Existing Data
[caption id="attachment_10800" align="aligncenter" width="604"] Image via smartinsights.com.[/caption]
The first thing to do is to analyze data from, for example, Google Analytics and see how your website works. And remember that there are factors at play that extend beyond design, so take a moment to consider them during this process as well. The chart above examines how a longer page load time can negatively impact the click rate of converted vs. non-converted customers.
Remember that factors like excess code and cookie storage can also impact your website visitors' decision-making process. Reduced conversion rates are not always a matter of design alone.
[caption id="attachment_10801" align="aligncenter" width="484"] Image via smartinsights.com[/caption]
Maybe you have a lot of viewers but very few buyers. Maybe people click on your products, but they don’t decide to buy anything. And maybe you simply haven't updated your tactics recently and customers have gotten bored with your old design.
These numbers will tell you what is happening, but they won’t tell you precisely what changes will increase customer engagement. One way to find this out is to make a survey. Because the "why" makes all the difference. The answer to the question why will also help you when you create your plan.
2. Plan Changes for Testing
Conduct a customer survey, then construct a a very detailed plan for what changes you need to make to your sales or opt-in pages.
You will make the changes incrementally, testing along the way to optimize the conversion on each element.
It is best to stick to one change at the time so you will know what actually made a difference. This doesn’t mean that you can’t make three different versions but keep to one change on each version.
[caption id="attachment_10803" align="aligncenter" width="826"] Image via optimizely.com.[/caption]
In the example above, a color change alone increased conversion rates. Of course, it is unlikely such a small change will make that big of a difference when you do your tests. Maybe your landing page doesn’t have any images, so you can test two different images to see which yields the highest conversion rate. From there, you would keep the image that better engages your audience.
From there, you might change the size or placement of the image.
Or you could see whether people are more likely to click on conversion buttons to the left or right of the page.
Or you can test to see the optimal number of social media widgets, as well as test which social media channels yield the most shares.
You should also figure out what you are trying to accomplish with these changes so you can compare your results to what you are doing at present.
“Make sure that before you start testing you have a clear idea of the results you’re looking for. You should already know your baseline result, which is the results you’re currently getting. You want to test option A and B against each other, but you also want to know that whichever one does better in the test is also doing better than your current results. Alternatively, you can use A as your control (leaving it whatever you’re currently using) and then use something new for B.”
-- A Beginners Guide to A/B Testing- An introduction, KISSmetrics
3. Make the Changes
[caption id="attachment_10805" align="aligncenter" width="568"] Image via LeadPages[/caption]
When you have made your plan, it is time to create the alternative version to your website. There are plenty of A/B testing software options out there, including Maxymiser, Optimizely, Qubit, VWO, and LeadPages.
Some are opensource and others you will have to pay for. Which one you decide to use is up to you. There is also the option of testing different software to see how well it performs.
4. Start the Testing
[caption id="attachment_10807" align="aligncenter" width="560"] Image via unbounce.com[/caption]
It is important to give the test some time. One of the things I clearly remember from my economic classes in college is that you can have hundreds of fancy theories and a surplus of statistics.
But you can rarely control or predict human behavior.
We have so many things that influence our behavior on any given day. The weather, our mood, what day of the month it is, if we are hungry or if it is close to the holidays. And this is the reason you need to give your test some time. We can prove almost anything with statistics if we misinterpret them. The point of A/B testing is to get clear results, so do not try to rush it.
5. Analyze Your Results
[caption id="attachment_10809" align="aligncenter" width="510"] Image via campaignmonitor.com.[/caption]
All A/B testing software will give you data that you can use to see how your changes played out. In the above example, Version A received over a third more clicks than Version B over a period of several hours. But as noted above, you need to give the test time. Results might be different at different times of the day, depending on who is logged in and who isn't.
“A/B testing is about knowing your visitors. You will learn valuable insights, because you will “know” instead of “thinking” about their preferences.” How to Build a strong A/B Testing Plan That Gets Results, Jaan M Sonberg, www.conversionxl.com/how-to-build-a-strong-ab-testing-plan-that-gets-results/
Maybe the alternative website you made drew more customers. Then you can implement the changes and enjoy more revenue.
If the changes you made didn’t produce any positive effect, you will have to start over again. Re-examine your designs and test a few different design elements until you see a positive difference.
“A/B testing can be an extremely important and helpful tool when it comes to maximizing your content marketing efforts. Don’t fall into the same old comfortable routines simply because they seem to work pretty well, you never know what variations to your strategy could be even more successful.”
Jonathan Mahon, The Importance of A/B Testing in Content Marketing, LinkedIn
A/B testing is quite simple in theory, but you may run into a few glitches when you are trying to implement these steps. It can be difficult to know how to change your overall design initially. This is especially true if you have never done A/B testing before.
If you still have trouble landing on the perfect heading copy, the perfect image, and the perfect design, don't hesitate to ask for help. Your landing pages are invaluable marketing tools. They should be allowed to reach their full conversion potential.