If you want your split tests to contribute to your bottom line, you need to know how to produce effective split testing ideas.
However, if you’re new to the game of split testing, you might not know how to develop such ideas.
For instance, you may not know what website elements you should even be focusing on. As well as that, you may also be unaware of how to come up with test ideas for each of the website elements you’re looking to improve.
Thankfully, developing split testing ideas is not as hard as it sounds.
In this article, we’re going to take a look at some of the steps you need to take in order to produce winning split testing ideas.
By the end of this article you should know how to identify split testing opportunities. You’ll also learn how to come up with alternative website elements that you can split test, after having found such opportunities.
1. Determine What You Want People To Do
Before you even consider split testing, it is important that you clarify what your end goals are. Defining a goal is often what separates success from failure.
Your split testing efforts are most likely aimed at encouraging more people to take a certain action.
For example, you might be interested in getting more people to sign up to your email list. Or it might be that you want an increased number of people to buy your product, once they hit the checkout page.
It is essential that you clarify what you want people to do more of. Knowing this information will help prioritize your split testing efforts, once you’ve later identified some opportunities.
If you want people to sign up to your email list, it makes no sense split testing to improve clicks that produce a different result.
This decision will guide your split testing efforts. If you don’t have an end goal, you can’t craft an effective strategy.
So take a moment to identify what metric you’re looking to improve before you deep dive into split testing.
2. Gather Data And Analyze The Current Situation
When it comes to formulating a split testing strategy, gathering data is important.
When we gather website data, we can identify split testing opportunities throughout a website. However, data also lets us capitalize on these opportunities by providing potential a/b split testing ideas.
So how exactly do we gather data and use it to produce split testing ideas?
Using Google Analytics to Gather Data
The first option is to install Google Analytics. With Google Analytics, we can see how people are behaving on our website as a whole.
For example, we can identify pages where people are consistently leaving the website, in disproportionate numbers.
We can also identify how many people click through from one webpage to another.
There’s a lot more you can infer from Google Analytics data, but by focusing just on those two data sets, you should have enough to get you going.
Let’s take the first example of being able to identify where people are leaving your website.
If you have a page that is causing a lot of people to leave your website, there is a chance that there is something ‘wrong’ with that page.
If you want to see the pages that are causing people to leave your site, do the following.
Click on ‘Exit Pages,’ to see which pages are causing people to leave your site
Login to your Google Analytics account, click on the sidebar and choose behavior. Then click on site content and choose exit pages.
Pay special attention to the ‘% Exit’ column.
The screenshot above was taken from a video shot by Neil Patel. It is from his free traffic university course. In it, he explains the concepts that we’re covering now.
The column that we’re going to be focusing on most is the % column. This will allow for you to analyze the data in context of people who click on that page and then leave.
As you can see, the arrow on the image points to a page that has an exit percentage of around 82%. This is disproportionate to the other pages that Google Analytics has collected data for.
If you see a similar result in your own Google Analytics account, you can come to the same conclusion as Neil — that something on that page is causing people to leave.
If you know that something is wrong with that page, and that something on it is causing people to leave, you now know that you need to make some changes on the page — a.k.a. you need to split test it.
If it were not for the data provided by Google Analytics you might not have been able to identify that page as a candidate for split testing.
In the second concept, we analyze how many clicks go from one page to another.
On your website, you may want to see if there is a page that causes a lot of people to click through to another page on the website.
By using the behavior flow section of Google Analytics, you can identify whether or not this is happening.
Below is a screenshot of what behavior flow looks like.
Behavior flow lets you know how people are moving through your site.
As you can observe, it allows for you to see how people are moving through your site.
Using this screen, you can identify whether or not pages are converting the way you want them to.
For instance, you may have a webpage in the ‘Starting pages’ section. When people reach this page, you may want them to click specifically on another page.
If everything goes as you want it to, the second page should appear in the ‘1st interaction section,’ with an acceptable number of visitors and a reasonable drop off rate.
This web page could be classed as a winning web page, because of the fact that it is keeping people on your site. It could also be argued that this page is converting well.
However, if the second page is nowhere to be seen in the ‘1st interaction’ column, you know that something is going wrong.
As a result, you therefore know that you need to split test this page.
Though how are you going to split test it? You know you need to, but what exactly can you do?
Remember, this screen also shows us ‘winning’ pages that are converting well.
Therefore, take a look at these winning pages and see if you can use what’s working for these pages on the poorly converting pages.
There may be a certain website element on that page that is causing people to click through to another page, as opposed to leaving the website.
If we can identify what this website element is, we can use it on the other web pages that are converting poorly.
So to wrap that up, Google Analytics can help you find web pages that need split testing, thanks to its ability to identify a webpage that is causing people to leave.
It can also help you to spot winning pages that are causing people to stay on the site and click through to other pages.
You can then emulate the traits of the winning web page on the web page that is causing people to leave. This should, therefore, reduce the amount of people leaving your site — of which has the potential to lead to a rise in revenue.
Using CrazyEgg To Help You Gather Data
CrazyEgg is a great way to gather granular data based on user experience.
Google Analytics is not the only way in which you can gather data. You also have the option of using something like CrazyEgg.
CrazyEgg will give you the chance to see how people are interacting with certain webpages on a granular level.
You’ll be able to see where people are clicking and how far down the page they are scrolling.
Of course, there’s a lot more data that CrazyEgg provides, though those two metrics should be helpful enough.
Let’s take a look at how data gathered using CrazyEgg can be used to identify split testing opportunities.
Let’s assume that you want more people to sign up to your email list, and that your email opt-in form is located on the bottom of your page.
Example: Using the Scroll Map
Here’s an example of a ‘Scroll map.’
With CrazyEgg, you’ll be able to identify whether or not people are in fact scrolling down far enough so that they can see the form.
Above you can see a scroll map. It shows exactly where people are scrolling to when they visit the page shown.
The white and red areas show where most people are scrolling to. The blue and green areas show where people are least scrolling to.
If CrazyEgg tells you that they are not scrolling to the bottom of the page, you then know that the opt-in form needs to be moved higher.
Of course the long term solution might involve creating a more compelling page so that people do, in fact, scroll to the bottom.
However, for the short term, moving the opt-in form higher, could result in an increased number of opt-ins.
Example: Click Maps
CrazyEgg ‘Click Maps’ let you know what’s being clicked on.
Above is an example of what a click map looks like in CrazyEgg. A click map is designed to show you where people are clicking.
The click map follows a similar pattern to the scroll map, in that the red and white areas show ‘high click’ locations and the blue and green show ‘low click’ locations.
A click map is helpful because it lets you know if people are clicking on the website elements you want them to.
Before we were talking about moving the opt-in form higher, as a result of the data provided by the scroll map.
Well once you’ve done that, you can then use the click map, to see if in fact, the location of the opt-in box was a problem.
If the opt-in box still has around the same amount of clicks on it, even after having been moved, you know there is a deeper problem.
Combining Google Analytics And CrazyEgg
What if your opt-in form is already above the fold, and still generates below desired of sign up numbers?
If that is the case, you then know that you need to create an opt-in form that is more appealing.
If you combine the Google Analytics data with what you learnt from CrazyEgg, this shouldn’t be too hard.
All you have to do is look at some of your more popular pages and then incorporate the themes of your winning pages into your opt-in forms.
For instance, your top page might be related to Facebook marketing — why not create a giveaway that is based on this topic?
It might also be helpful to use some of the same verbiage that is on your winning pages, within your opt-in forms.
If people are already resonating with that style of content, there’s a chance that it could improve your opt-in rates if used strategically.
As with all things, let the test run and then see what the data tells you.
3. Watch Someone Use Your Website
Watch in person how people are using your website.
Watching someone use your website can help unearth some potential split testing ideas.
The best way to do this is by getting someone to take a specific action — i.e. ask them to buy a product on your site.
Be careful, as you don’t want to include any bias in the situation and influence their decision.
But by watching them go through the process of buying a product, you’ll be able to spot some of the friction points they experienced throughout the entire process.
Once you’ve identified these friction points, you can then remove them by split testing some alternative solutions.
Of course if you don’t know what these alternative solutions could be, consider asking the person you watched used your website.
This person should be able to tell you what their pain points were and what you need to do to in order to fix them.
How To Find Someone Who Can Trial Your Website
If you’re struggling to find someone nearby, of whom you can watch use your website, you might want to use a service called ‘Peek.’
Their paid offering, allows for you to set tasks and then receive video recordings based on how the whole thing went.
This can be a handy way to get the job done, if you were struggling initially.
Alternative: Consider Using A Survey
You can use surveys to learn more about the user experience and any potential friction points.
Another way you can learn about the user experience, is by asking questions through the use of a survey.
With tools such as Qualaroo, you can now use implement something known as ‘exit surveys’ onto your site.
These surveys will appear when a visitor chooses to leave a certain page.
Let’s imagine that you’re running an ecommerce store.
If you set it up so that these surveys appear on checkout pages, you may be able to reduce cart abandonment rates.
That is because you can create a survey that will ask the visitor, why they’re leaving your site, especially this late in the buying process.
You might have list of ‘tick box options’ that visitors can click on. These options will list potential reasons for leaving your site. Tick box options could include:
- Shipping cost was too high
- I found it cheaper elsewhere
- Need to do more research
- The website didn’t look trustworthy
- Too many details asked for/Checkout process overwhelming
- Didn’t accept PayPal
- No information on delivery times
The questions you ask will depend on your niche, as those used above relate mostly to an e-commerce store.
If you’re running a SAAS company you might put:
- No free trial
- Competitor offers better product
- Not cloud based
- Not trustworthy site
- Monthly cost too high/Could not justify
- Need to speak to someone before buying
Of course you don’t have to use tick box options only, as you may also include a text field, where people can enter more information.
Alternatively, you could choose to send people a survey after they’ve bought something from you.
In this survey, you could ask questions which reveal why they decided to buy something from you.
When asking these questions, you may just want to flip the questions you used on the exit survey. If running a ecommerce store you may for instance ask –
‘Why did you decide to make a purchase with us?’
- Shipping was cheap
- Was the cheapest site
- Clear information on delivery times
- Website looked trustworthy
- Didn’t ask for too many details/Checkout process was easy
- and so on…
And for a SAAS?
- Free trial was offered
- Was able to speak with someone before purchasing
- Best product available
- Site looks trustworthy
- Cost effective
- Was cloud based
- and so on…
If people keep on telling you that they decided to buy from you because you provided cheap shipping or because of a free trial, you might want to emphasize this point elsewhere on your website.
In doing so, you may just increase your conversions, as you give people a reason to buy from you and conquer one of their main objections early on in the process.
4. Borrowing Ideas
When it comes to producing split testing ideas, the points we’ve mentioned can seem quite complex to a newbie.
Their complexity is worth it though, as there’s a chance that they’re more likely to produce results, since they’re tailored to your website specifically.
Nevertheless, you don’t have to always look within your own website, when it trying to develop new split testing ideas. You could also look at other websites to see what is already working for them.
For instance, if you see a company that has a landing page style you like, you may want emulate it, to see how well it works for you.
You may even choose to borrow another websites headline strategy, in order to produce some effective split testing ideas for your headlines.
Though as you might have guessed, you need to be careful when emulating ideas.
That is because when you’re borrowing elements from other websites, you have to account for the fact that they may follow a different business model. As well as that, their target market may respond to things differently when compared to yours.
Keep in mind, however, that a solution to this problem is to emulate the concepts you see and implement them using data based on your own website.
Let’s take the examples of split testing the use of content upgrades in order to boost subscriber numbers, after having seen the strategy successfully implemented elsewhere.
In such a situation it helps to identify your top five posts and create content upgrades for each of these posts. You can use Google Analytics to do this. You’ll then be able to quickly tell how well this strategy works for your own website.
By using data to implement foreign ideas, you improve the chances of success.
5. Go against Best Practice
Best practice is when you do something that seems logical because everybody else does. You could argue that it’s when a certain style becomes the standard approach.
We’ve not talked much about best practice in this post, as it was assumed that a lot of your tests were based on best practice.
However, when it comes to split testing, sometimes best practice can be your worst enemy. People might get used to a certain style of landing page, and hence react less and less to it.
A certain landing page might be considered best practice for one market, but not best practice for your own target market. Certain landing pages, have in fact, benefited from being unconventional.
The point being, is that you should always experiment with some radical ideas that do away with what is ‘supposed’ to work. As logic dictates, some of these tests are going to fall flat on their face within hours of being implemented.
Some, however, might just surprise you and showcase that best practice does not count for as much as you thought it would.
A good way to do this is to ask what might happen if you did the opposite of what you’re currently doing.
Time To Implement
For many of us, having epiphanies based on what we need to be split testing is not a daily occurrence. Yet by using some of the frameworks provided above, we don’t have to rely on epiphanies any more.
Rather we can hunt down split testing opportunities using data based on what current website users are already doing. And if that’s not enough you can even ask website users what needs improving by using a survey
Once you’re doing those things, you can even consider the option of scrapping all that you’ve learnt to try something you’ve never previously considered.
And whilst that might sound a little nihilistic, there’s a chance that it could bring you results you never thought were possible.
Whatever path you decide to follow, the most important thing is that you get going. Implement some of the ideas suggested and see how well they work for your site.
Leave a comment if you enjoyed this article or if you have any questions.