A/B Testing (© tashatuvango / fotolia.com)
A/B Testing (© tashatuvango / fotolia.com)

Split testing is a powerful tool that businesses use regularly in order to test their products, their websites, their sales pages and more. This process can be used to ‘evolve’ any website to a point where it will become optimized for users and thereby guarantee success.

The more you read about business, the more you realise that it's a simple matter of applying systems that have been proven to guide profitable decisions. Using these systems, you almost reach the point where the product or the service doesn't matter quite so much as the business model you use to sell it. And a fine example of this is split testing, or A/B testing, which is a business tool that allows you to essentially 'evolve' an idea to the point where the market will be guaranteed to accept it. Let's take a look.

The Concept: What is Split Testing?

Essentially the idea of split testing is to take advantage of the fluid nature of digital offerings by gradually tweaking them and looking at the results, but with the added caveat of only letting a portion of your market test those tweaks.

The reason it's called 'A/B Testing' sometimes then is that you are going to be coming up with two versions of your product or service which you could call the 'A' version and the 'B' version. The A version is the baseline version that you've already released, while the B version will be the same thing but with a slight tweak and will be released to only say 10% of your market. 

What this then allows you to do is to test the performance of your product or service with those changes in place and to then decide whether you want to stick with the changes across the board or not.

An Example of Split Testing

So, let's take this here website as an example. Let’s imagine that I have some AdSense on a webpage which is designed to make me money, and it's in the top right at the moment because I believe that's where it will make me the most money. But how can I be sure that moving it to the middle wouldn't make me more money still?

Well I can't, which is why I might want to test that theory. Only if I moved it to the middle and I didn't get any clicks then I wouldn't know for sure whether that was because of the new position, or whether it was because of some other factor (maybe I moved the ad during the Strictly Come Dancing finale and no one was online). At the same time I'd be risking losing my current audience - you guys - because the ad might be so offensive in the middle that you'd leave and never come back.

So, to take full advantage of split testing what I'd do instead is to write a PHP script that would redirect people from Wales to a mirrored version of this site and then I'd look at the statistics and I'd be able to see whether or not moving the ad was a good idea. There'd be no confounding variables other than the fact that the people on the B site were Welsh (we'll assume Welsh clicking behaviour is the same) and if they hated what they saw then I'd only lose some of my Welsh contingent and the majority of my traffic would be none the wiser.

This then means that I could find the optimum position for my ads, and I could find the optimum font, the optimum colour scheme and everything else for that matter by looking at cold hard numbers. And I could do the same with my apps too and anything else that I wanted to be a

success. In short split testing allows you to practically 'evolve' your website or software or anything else to suit the market it's aimed at. It's time to come down from the trees!

Split Testing vs Focus Groups

So the idea of A/B testing, is to allow you to test an idea without having to fully commit to it. For instance then, if you were to create a new web design then rather than release it completely you could instead test the water to see if the market was receptive before you committed to the new look.

The question of course is how you go about doing this. For big businesses that would mean focus groups and market research, but unfortunately for many smaller start-ups or entrepreneurs these resources just aren't available. 

Where split testing differs then is that you use a portion of your existing market to test your changes. This way for example you could use a PHP script to direct only a small portion of your traffic to your new look website. This might mean for instance that you send people from a certain area there, or that you only send people there during a certain time.

This way, you will be showing only a small portion of your visitors the new site, and that will allow you to test the reaction and compare it to the performance of your original set up. You could look at your click through rate for instance then, and if it were to double for your site redesign then you'd know to stick with it; while conversely you'd know to dump it if you suddenly lost clicks.

This gives you the great advantage of being able to look at the statistics for your old and new sites side by side, and at the same time it also enables you to test the changes without losing your entire audience. If your site is hideous enough to turn people away, then at least you'll only damage your reputation with a small portion of your traffic.

Getting the Most From the Process

To get the most from this you should focus on small iterative changes. In other words you shouldn't attempt sweeping redesigns because you will be committed to those once you've gone to the trouble of creating them and because it will be harder to pin-point which areas have been effective or not. Rather, try making little tweaks to your colour scheme, your ad placement, or your font, and then use a split test to find the perfect choice for each.

A Word of Caution

In theory, you can use split testing in order to hone and evolve any web design or any product into the perfect offering. You are literally seeing what works and then doing more of that so in principle, this should be fool-proof.

But the operative phrase here is ‘in principle’. In practice, it is very possible for things to go wrong and the reason for this is that you don’t have a big enough data set.

Really, a split test is very similar to the way that researchers carry out studies in the lab. The big difference is that a researcher will make sure that they have very large groups of participants that have been matched to members of the general public in terms of their demographics. They’ll eliminate outliers and they’ll perform statistical tests of significance.

The point? Your study doesn’t have a large enough group and neither is that group controlled enough. There is still a chance that what you’re seeing is coincidence. It’s a small chance true, but just be aware that even the results of a split test are not gospel. What’s more, is that the longer you run the test, the more useful the resultant information will be.

More Ways to Use Analytics in Your Web Design

Split testing is an example of using web analytics and stats in order to measure the objective successfulness of a website. There are many other methods for doing this however, which work in a similar but distinct manner. Many of these are accessible through Google Analytics.

Google Analytics is an incredibly powerful tool for any webmaster or blogger who wants to increase the quality and quantity of traffic to their website. Providing reams of data regarding the nature and volume of traffic coming to your site, the possible ways you can interpret and utilise this information is almost endless.

You can use analytics then to not only see how many people are coming to your site or clicking on your links, but also where those people are from, how long they’re spending on your site, which pages they are reading in particular etc. This information then allows you to redesign your site to fix what isn’t working and to gain a better understanding of who visits your site and why. This goes perfectly hand-in-hand with split testing.

Tips for Smart Use of Data

Here are some tips that will help you to make better use of Google Analytics, split testing and your metrics in general:

Measure Success Using a Single Metric

Tim Ferriss, productivity guru and author of The 4 Hour Workweek, recommends that whether you're dieting or promoting a business online, it can be a mistake to focus on too many different data points when measuring your progress. When you visit Google Analytics you will be presented with reams of data, and deciding what to prioritise can be confusing ultimately hampering your progress rather than aiding it.

The rest of this data is highly important and can be used in many ways. But in terms of setting a single goal, it's a good idea to choose one metric to try and increase and to focus on that. This is simple, straightforward and allows you to view your success through a single lens and make decisions based on that. A good one to start with is your 'bounce rate' by looking at the time people spend on your site. Get people to stick around longer and you'll find other factors like your number of unique visits/conversions take care of themselves.

Use the Goals Area

Luckily Google seem to agree with this strategy and have thus provided an area of Analytics specifically designed to allow you to pursue specific goals. Configure this goals area and you can test outcome-oriented metrics that matter to your site.

Carry Out Tests

Using the data provided by Google Analytics should be a two-way affair, meaning that you should be interacting with that information and seeing what you can do to effect it.

Split testing is just one example of these sorts of tests. You can create a range of other tests by tweaking aspects of your marketing for instance to see how that is reflected in the data.

Follow the Money

Lord Allan Sugar is constantly telling his Apprentice contestants that they need to ‘follow the money’. That means that they need to smell what’s hot and then follow suit: meet that demand.

This is something you can do when using analytics to hone your business. So instead of trying to come up with an idea that will bring people to your website and then optimize your site for that with split testing etc., you instead look to see which searches are bringing the most people to your site, which demographics are enjoying your content etc. You already have this audience, so now you can just double down on what is working and ignore the rest!

Learn Your Way Around and Invest Some Time

You should spend some time learning your way around to find all the different features available and you should invest some time setting everything up. Google Analytics allows you to create custom dashboards for instance that show you all the data that you deem valuable so that it's available at a glance. A little work up-front will save you a ton of time later on.

You can also benefit from looking into some other tools. Google Analytics has an app for instance that lets you easily check your stats on the move, meanwhile a tool called 'Cyfe' can let you see multiple dashboards for multiple websites all in one place. If you have a YouTube account then make sure to check YouTube's own analytics page for more useful data.

So, spend some time getting used to Analytics and learning your way around. The guaranteed return on investment will make it more than worth your while; in this case knowledge really is power.

Creating a Great Landing Page

One of the most common uses of split testing and analytics, is to create a landing page or a sales page. This is a page on a website that is designed entirely with the aim of selling a product or service and the entire design is built around that fact.

Split testing is perfect for this because you have a very clear end goal: getting people to click ‘Buy’. And the more you try to hone your design, the more you can drive your conversion rate up.

Here are some more tips for your sales page:

Use a Funnel Design

When you create your ecommerce site or landing page, it's important to make sure that you keep the traffic on that page first and foremost and that you prevent them from leaving however you can. The way most sites accomplish this is simply to keep things barren - run a long and narrow passage of sales text down the middle of the page and avoid any links pointing to other pages or sites. That way, once someone is on the right page, they will naturally scroll down to read more and they won't be tempted to visit anywhere else.

Use Headings Well

Headings are incredibly important in your site design because they provide your visitors with a way to quickly skim through a large body of text. If you are concerned that your content might be a bit heavy going, then use separate headings so that people can quickly get the gist of it without having to read it all fully.

Another important tip here is to tell your story in your headings. This way, even if your visitors only read the headings - which a lot of people will do - you will still convince them that your product is at least worthy of further investigation.

Avoid Break Points

Break points are points in your text where you risk losing your readers. If you have passages of text where you begin to ramble, or large physical gaps on the page with nothing interesting in them, then you risk drawing your readers back out of what they're reading and losing their interest. It's important to make sure that your text has a rhythm and that your text is organized in a way that will keep viewers scrolling down the page and moving from one point to the next. To test for break points, read the text yourself and make a note of any points where your mind or your mouse starts to wander.

There is much more to consider, but when employing any of these changes, just be sure to check in with a quick split test to see if it is working for you!



Press releases


Search