A/B Testing: Improve Your Conversion Rates

What is A/B Testing?

Put simply, ab testing is the comparison of two examples of copy to determine which is more appealing to the consumer. We will talk about the types of copy that A/B testing can be applied to later, but now let’s try and expand on the above with a more understandable example.

A/B Testing involves creating two samples of the copy. Let’s imagine we’re talking about a website landing page. You need that page to be attention-grabbing, to give the reader the information they need, and to encourage them to move forward with the transaction.

A/B Testing uses a control version – A – and variant – B – each of which is measured against your key metrics.

Say you have a page that is not converting at the same level as others. There will be a reason that this is happening. It could be that the copy is not relevant or poorly written, it may be the images are not informative to a level that is acceptable. It could even be a combination of colors.

By creating a new version with a different copy, for example, you will be able to measure the change in performance. That’s the basics, but we want to go a little deeper into how – and also why – split testing works, and when you should use this effective tactic.

When Should You Perform A/B Testing?

The simple answer to the above is ‘always.’ Split testing works best when it is continuous. This may sound like an arduous task but in fact, it can be easily implemented. A/B Testing is about fine-tuning. No website, email, newsletter, or ad is ever perfect, and each can be improved. Continual A/B Testing is a way of finding the small improvements that can make all the difference.

The bonus to A/B Testing is that there are so many areas of your marketing campaign that can be used to analyze. Here’s a list of some of the most obvious, which we will look at in more detail further on:

This is by no means a comprehensive list but is a simple example of where your A/B Testing routines can be implemented. Split testing allows you to iron out problems in your website and other marketing media. It can show you where improvements can be made, and also where you are using marketing campaigns that are having little or zero effect. It enables you to understand what your target audience is interested in, what drives them to move forward, and where you can utilize your marketing budget and efforts to the best effect.

To help a little more, here are a couple of examples where you would use multivariate testing to ensure you get the best results.

Example 1:

You are currently running a digital marketing campaign that is not hitting expected targets. Perhaps you are not seeing the conversions you expect from traffic, or maybe consumers are visiting a certain page and not going any further.

By creating a variant of the current page – change the text, for example – you can measure and see if that is the problem. If there are no changes with the text, change the image, and so on. You will find a part – or parts – of the page that is the problem. This can be performed across all your marketing media and will provide results.

Example 2:

You have a launch planned – a product, perhaps – that you are looking to put your effort into. You need to determine which method of getting your consumers interested is best. Is it via messaging, or email? Would social media be the most efficient method? Or will the customer respond to a newsletter more readily?

By running one method at the same time with another in an A/B Testing routine you can measure the metrics and results that each version provides, and launch using the most effective option.

We will be looking at the main benefits of multivariate testing shortly, but first, we want to talk about the importance of accurate analysis and the key metrics you need to look for when running an A/B Testing campaign.

A/B Testing Metrics

Now we understand what A/B Testing is all about, but what are you actually looking for? What are the key metrics that are important to your analysis? That depends entirely on your business and the medium that you are performing your multivariate testing on. Using a couple of examples here’s what you should be looking at:

Those two examples should give an idea of the key metrics that drive your website or email success. We need to look at the analysis in more detail, as it is perhaps the most important aspect of an A/B Testing routine.

Analytics in Developing an A/B Testing Project

Before you even think about beginning an A/B Testing routine you need to have a benchmark to work to. This is your current performance in terms of the data you are testing. So, the first thing to do is to collect data. You will already have this information via your data analysis tool such as Google Analytics, and we recommend the data you use is absolutely up to date when you begin the testing. If you don’t have knowledge of analytics, it is important you bring on board someone who does.

For example, you may be a great salesperson but leave the analysis side to others on your team. Now is the time for them to use their skills. So, we know that A/B Testing is about comparing a control version of the copy for comparison – this could be your original that is underperforming or a new alternative – against a variant that you have created especially for this test.

There are some important key data points that you need to have in place from your original information. These form the background of the A/B Testing routine, and among the most important are the following:

The above are the bedrock of a successful A/B Testing routine as they form the standard that all other examples will be compared to.

Many people make the mistake of basing their analysis on personal impressions. It is essential that an A/B Testing project is based on gathered data of the user behavior, even if that data goes against a hunch or prior belief. We cannot stress how important it is to begin with a clarified and reliable data starting point.

The gathering and analysis of the above key data points is your preparation for the test, and only once you have this information ready and available can the test itself begin.

Before we move on to explain the best ways of analyzing your test results and how to use them, let’s talk about how to run the test itself.

Performing the A/B Testing Routine

There are nine basic steps involved in designing and implementing multivariate testing routines. To ensure that you get reliable data you need to stick closely to these rules. They are as follows, and explained as simply as possible:

What you should take away from this test is a clear indication of the problem area that your control page or copy – or the variant – is presenting. A/B Testing is not a difficult process and the more you use it, the better you will become at creating that all-important hypothesis that defines the consumer experience. You will take away clear evidence of what you need to do to improve performance.

As we have already mentioned, A/B testing is not a one-off procedure. It is something that needs to be carried out throughout the lifetime of your digital marketing projects. It will show you constant updates you can make to your marketing.

For example, if emails are not performing and SMS messages are, you know where you need to make changes. Remember to trust the data; don’t rely on marketing gurus who tell you to go in another direction, and don’t base your changes on a hunch.

Briefly, we need to jump back to the points about analytics, as this is an area that needs your continued attention during the test.

Analytics During the A/B Testing Project

The best information you will have available is that which comes to you in real-time during the A/B Testing project. This is why you need to keep a close eye on what is happening in front of your eyes. You can use one of many dashboard tools that will keep you updated with rolling results. This means that you can see current trends and take into account different times of day, for example, where consumers may act differently.

It may be that your test begins returning results that are clearly anomalies. In this case, there is obviously something wrong with the format of the test and it should be revised. You might start seeing clear trends towards one angle in the early stages of the test. It may be that at this point you want to amend the test to reflect this.

An A/B Testing project is fluid and may change at any time, and this real-time information is of vital importance in your eventual analysis. The final result may not be returned by the first attempt. For example, you may find that neither the control nor the variant versions met with favorable results. In this case, you need to create a new copy that is notably different from that you have used. Next, we want to have a look at how best to interpret your A/B test results as this is the key to the decisions you make.

How to Interpret A/B Test Results?

As we have seen above it is vital that an A/B Testing routine is run to a strict framework. In addition to what we have already talked about, you need to go into the test with a set goal as to what you wish to achieve. You need to measure one element of your site over the others so that this element can be fine-tuned to optimize performance. We’ll look in more detail at the usual areas that are analyzed in A/B Testing but to help, here are a few things you may wish to investigate:
The number of visitors to your site, page, or store.
Open rates for emails and messages.
CTA click-through rates.
Signups for newsletters and marketing materials.
Subscriptions to your store or site.

We should say that A/B testing can be used as an analysis for all or any of the above, and we have a bigger list to talk about further on.

By monitoring the results during the test – the real-time element we mentioned above – your dashboard will soon begin to show a notable statistical result. You need to set a time period during which this result can be determined. It may be that the A email, for example, gets a 40% click-through and the B email just 20%. This is a notable difference that you can use a guideline.

It’s important to know that when dealing with conversion rates the data can be measured not only in numbers but also in total revenue. This is important data when, for example, analyzing the spend variance over different product ranges. Furthermore, conversion rates can also be measured as part of any action and not just sales. For example, leads generated, clicks on ads, and time spent on the site are all valuable metrics.

Let’s expand on the above list to look at the various metrics and areas of your digital marketing that can be tested in this way.

What Can Be A/B Tested?

As we keep reiterating throughout this article, there is a great deal of elements of your digital marketing that can be successfully tested using A/B testing.

Let’s have a look briefly at some of the most important elements of your page, site, or store that can benefit from A/B testing:

Navigation links are an important part of any website or store and should enable your website visitors to navigate around the site easily. One common reason for leaving a site is poor or confusing navigation. The modern consumer expects simplicity and efficiency throughout the online retail process. Running a split URL test regularly should give you an idea of the frequency of use and of dropouts because of them,

Calls to action (CTAs) are among the most important parts of your written copy. This is the sentence or two that is designed to encourage the consumer to move forward and complete the transaction. A poor CTA will have a negative effect on your ROI. Test power words regularly by A/B Testing to see what tone or style appeals to your customer base and amend where needed. This is one example where a large number of subjects is essential to get that statistical significance.

Landing Pages are where the consumer ends up when they click on an ad or search but are they reaching the landing page and going no further? If so, you have a problem with your page as it is not offering enough for the consumer to take things further. It could be a problem with the written copy, or perhaps the images involved, and you need to keep on top of landing pages on which visitors spend an excessive amount of time. Run A/B Testing regularly and try different approaches with the copy involved.

The design and layout of your site or pages are an essential part of the engagement process. If a design is bland and unattractive it will drive consumers away. Likewise, if it is too busy and cluttered it will have a negative effect. The consumer likes to see a clear and understandable design and a layout that allows them to understand what is on offer. If your site is more than a few years old, it will begin to look outdated. A great proportion of consumers find old-looking and drab designs off-putting.

Copy is not just the written word but all your content on your site or store. Copy needs to be spot on in terms of quality and relevance, and you need fresh copy regularly. If you use A/B Testing on the copy that has been on the site for some time you will see a drop-off in engagement. This is natural as the consumer wants to see new and fresh information. Try A/B Testing with, for example, simple and informative copy as the A version against more humorous and lightweight copy on the B version. You may find an original approach appeals more widely.

Pop-Ups are a great marketing tool but are yours getting the attention and response you need? It’s easy for consumers to ignore pop-ups so you need to have an approach that appeals to and engages them. Run a version that simply explains an offer against one that offers a discount, say, and see what the difference is. Keep a check on pop-ups and their content as this is one of the areas where A/B Testing can make a great difference.

Ads are similar to pop-ups in that they can readily be ignored. Your ads may not be grabbing the attention, so try something entirely different and see if that hooks your fish. A/B Testing is essential with ads if you are designing a major campaign as a poorly designed ad can be the death of a product launch. Use A/B Testing regularly and thoroughly before launching ads to a wider audience.

Headlines are there to get attention, but are yours succeeding? A/B Testing will tell you yes or no when you run your standard headline against a more striking example. The headline is the only thing a consumer will read if it does not interest them, so what interests your target audience? Try a few different approaches to headline writing and see what results you can gather.

Product descriptions need to be accurate, catchy, and informative. Often, a store will find that the description is not giving the reader enough information. Other times it is too long, and the reader moves on. This is a perfect category for multivariate testing.

Emails remain among the most successful marketing mediums, but you need to ensure your email title is enough to inspire the recipient to open it. This is among the most-used areas of ab testing as you must get the title and subject line correct for your customer base. Spend plenty of time on this subject and do the same with SMS messages and any other messaging formats you use.

Images are an important part of your marketing but are those you are using attracting attention? Poor images that are not high quality will not elicit a response. Those that do not show the product correctly will be ignored. Test which types of images get the best response by using A/B Testing as this is an area in which it is vital to keep ahead of the competition.

Social media buttons need to display a good click-through rate, otherwise, there is something wrong with them. It could be you are using the wrong media most times, and the benefit of a split test putting your Facebook buttons against Instagram, for example, cannot be overlooked. Finding the right social media marketing strategy is essential, and A/B Testing can help a great deal in this area.

Logos, taglines, and slogans are there to reinforce your brand, and if they are not getting the clicks you expect then a split test should help you find out the best way to improve them.

The above is not a comprehensive list, but it should give you a good idea of the sheer scope for improvement that a split test brings about. Before we finish, we’d like to run through some of the aims and benefits that you are using A/B testing to achieve, so let’s take a look at the main areas in which your business can gain from A/B testing.

Benefits of A/B Testing

Now that we know what A/B testing all is about, we need to have a quick look at the various benefits that you will gain from running well-designed tests, so here are some of the most important.

Solve visitor pain points with A/B Testing by analyzing user behavior at different points during the consumer journey. You can then redesign any of the elements you have tested to appeal to the consumer more effectively.

Get better ROI from existing website traffic by constantly running A/B Testing projects to fine-tune any areas that may be showing a drop-off in consumer interest.

Reduce bounce rate by analyzing content and other areas of the site or store in more detail. A/B Testing will show you which copy appeals to the consumer and which does not, so you can amend as needed.

Make low-risk changes to your copy, store, or site that enable you to improve the customer journey. By using a sensibly sized number of test subjects you can ensure that a relevant and suitable statistical return is enabled and reduce the chance of making the wrong decision.

Redesigning your website to increase conversion rates is perhaps the ultimate aim of A/B Testing although by using this method correctly a complete and costly redesign may not be needed. A/B Testing can show you where you can change small features and achieve greater conversion rates when it is used with a sensible and accurate analysis.

The benefits of A/B Testing go much deeper than the above points as we hope we have shown in this article, but what is certain is that this is a vital ongoing routine for any digital marketing campaign that aims to be successful. Before we close, a few tips on mistakes to avoid when designing and running your A/B Testing projects.

A/B Testing Challenges and Mistakes to Avoid

We’ve covered the subject of A/B testing, how to design and perform A/B testing, and what the results should give you. Now we want to run through some of the common mistakes people make when they are new to A/B testing so that you can avoid them. We’ve put these mistakes into a series of classes, so let’s begin by looking at mistakes made when deciding what to test.

A/B Testing Mistakes when Planning the Test

1: Focus on the Main Events

It is not possible to test everything. Therefore, you should focus on the parts of your site, store, or page that have the greatest impact on the customer. Consider these:

Subject lines and whether they work best as the short or long-form, what the tone they carry can be, and urgency. Also, look at personalization and calls to action and see if they can be improved. Headlines, hero images, and other factors are also impactful on the consumer and should be tested.

2: Don’t Forget Your Target Audience

This is vital advice as all too often copy is presented in a form that might not appeal to the demographic of your store, site, or page. Make sure you are writing for the correct audience. An A/B test is a great way of coming to the right conclusion here.

3: Establish Clear A/B Testing Goals

We’ve mentioned the importance of goals a few times but it’s worth mentioning again. Your A/B test will go nowhere if you have not set down an aim. It needs to be directed at one particular metric to arrive at the correct result. Don’t try and test more than one item at a time as you will return confusing and inaccurate results. Take the time to plan the goals individually and stick to them. Basically, are you running the test to learn, or to win?

4: Use a Large Enough Test Audience to Reach a Statistical Conclusion

Many marketers make the mistake of using only a small selection of consumers in A/B Testing. You need to include enough to perform a statistical analysis and arrive at a conclusion. What you are aiming for is 95% proof that you have a result that is not mere chance. Clearly, you need a large number of people to achieve this successfully.

5: Always Have One A/B Test on the Go

A/B Testing is the key to optimizing every area of your website yet as you can only test one element at a time it needs to be ongoing. Draw up a reusable template for results analysis and when one A/B test ends, start another with a different metric in mind. This way you are constantly modifying your site or store to be in line with what the consumer wants.

The above are just five of the common areas in which mistakes are made in A/B testing. We should add that as you get more into the process you will begin to improve your tests and therefore see more relevant results. The key to A/B testing lies in knowing what you want to discover and treating the unexpected with advanced and detailed analysis. Remember, no marketing campaign is perfect, but with a well-developed A/B testing routine you can be altering the faults easily and quickly and improve your conversion rates.

Final Words

We believe that we have covered the subject of A/B Testing as comprehensively as possible in the above and shown you the importance of this method of analyzing digital marketing performance. By going into the process with a definite goal and aim you are ensuring that the results you see are those that are as accurate as is possible.

If you follow the tips we have given about designing and developing your A/B Testing routine we are sure that you will begin to see the benefits very quickly. Remember to test only one item at a time and start with the most important, and you will see the changes that can be made from that single element of your campaign.

The change may be small, but the aim of A/B Testing is to ensure that your store or site is constantly improved and optimized for your customer base and that the consumers who use your brand enjoy the best possible consumer journey.

Start now by looking at existing data and finding one area of your campaign that is not getting the results you expected and put it to A/B Testing. We hope we have helped you understand this subject more readily so you can improve your digital marketing.