Many websites reach a stage of maturity where you have already overcome the obvious usability hurdles and applied the most common conversion optimisation techniques.
Although this is an excellent place to be in, it does represent a challenge when seeking to improve conversion further.
The danger is that those responsible for the website will start implementing changes to the site without having a clear idea of whether those changes will have a positive impact or potentially make things worse.
Each change will take time and money to implement, without anything other than a gut feeling about whether it will help.
To avoid wasted effort and the chance of lowering the conversion rate, I would recommend implementing a process to identify, assess and trial possible site improvements.
This process will need to combine an analysis of the data with some intuition to identify and fix problems. However, with experience, you will become better at it.
That approach should begin by identifying problem areas.
Identify Problem Areas
Websites provide us with unprecedented amounts of data about user behaviour, and we can use that to identify potential problems with a website.
An excellent place to start is with Google Analytics (or whatever analytics package you use). In particular, you want to pay attention to the top exit pages. These are likely to indicate where things might be going wrong.
Be careful, though. You are interested in the top exit pages as a percentage of total visits to that page. Some pages will get a lot more traffic and so will naturally have a high dropout rate. You are interested in pages that have a high dropout for the number of visitors it receives and is an important page in the user’s journey towards taking action.
In most cases, there may be something on those pages which are putting people off, or they are lacking something that people need to make a decision.
Two things to pay attention to with these pages is “average time on page” and the journey that leads to the page.
“Average time on page” will give you an indication of whether there is something on the page that is putting them off or something missing. If the “average time on page” is high when related to other pages, it might be that the user was searching for some piece of information they could not find. If the “average time on page” is low, it could indicate that something on the page put them off as soon as they spotted it.
By paying attention to the previous pages the user looked at before exiting the site, you may also be able to pick up trends. These trends might indicate that the journey is broken in some way.
This approach is not an exact science. It takes intuition to see where a problem might be. However, paying attention to where people drop out of the site will prove insightful.
Once you have identified a problem page, the next step is to establish what exactly is going wrong on that page.
Diagnose the Exact Problem
An excellent place to start when it comes to diagnosing the exact problem on a page is to look at session recordings.
View User Sessions
If you don’t have these already, session recordings are video recordings of somebody navigating your website, and you can obtain them using a tool like Fullstory or Hotjar.
Filter your session recordings to only include visitors who visited the particular page you are concerned about and watch what they do.
In many cases, you will be able to observe a user struggling to do something, such as complete a form. In other cases, you will see them skip over a crucial piece of information or fail to scroll far enough down the page.
However, sometimes, you will not observe anything that seems to indicate the issue with the page. In those situations, you need to turn to usability testing.
Run Usability Testing
Usability testing does not need to be time consuming or expensive. These days there are a wealth of services out there that will allow you to test your site with a handful of users for just a few dollars.
The first step is to identify a task you can ask users to complete that will involve them visiting the page that you are concerned over.
Next, you signup for a service like UserBrain and tell them the task you have devised, as well as how many people you wish to test.
They then go away and recruit users based on your specifications, and those users record a video of themselves completing the task. As they complete the task, they talk about their experience so that you know their thought processes and what problems they encounter.
In most cases, this will give you a pretty good idea of what the problem might be. However, if you are still struggling, you can simply asking people.
Run a One Question Survey
Often the most straightforward approach to understanding why users are not converting is to ask them. A one-question survey shown to users just as they go to leave a page (known as an exit-intent survey) can be a useful way of getting to the root of the problem.
I typically ask a question such as:
If you decide not to buy today, it would be useful to know why.
And then offer them a multiple-choice list of possible answers as well as a field to enter a reason not listed.
Together a survey, usability testing and session recordings should give you a good sense of what the problem is. The next challenge is coming up with a fix.
Trial Potential Solutions
Coming up with ideas of potential ways of improving conversion once you know the problem is not hard. What is harder is being sure that you have come up with the best fix possible. For that, we have two possible approaches; usability or A/B testing. Which method you adopt depends on the complexity of the improvements.
A/B Testing
If you can quickly and easily make the changes (such as changing copy or styling a button differently), then the best approach is to test this with A/B testing.
A/B testing allows you to compare multiple approaches and ensure you pick the best possible solution while avoiding making things worse.
In case you are not familiar with how A/B testing works, you show a percentage of users an alternative version of a page with whatever changes to copy, imagery or styling you have made.
You can compare as many different versions of the page as you want by directing different users to different versions and see which converts the best.
A/B testing can be set up and run using a free tool like Google Optimise, and you can see results in realtime, although to declare a reliable winner will take longer.
Where A/B testing cannot help is when you need to make more substantial changes to a page, like significant layout alterations or changes to functionality.
In such cases, it would be cost-prohibitive to build multiple versions and test them on the live site.
Instead, in such situations, it is better to prototype the changes and run a usability test.
If the changes involve making alterations to multiple pages, you can then run another usability test just as I outlined above. However, if it is a single page mockup, we have to adopt a different approach.
Because we have produced a mockup of the changes to a page and not built a working prototype, we cannot ask users to complete a task. What we can do instead is to complete one of three alternative tests depending on what changes you have made.
Preference Test
The most basic test is to ask people to compare the old version with the new version and select the one they prefer. This test can be useful when you have made aesthetic changes to the site or adopted a different tone of voice.
First Click Test
A first click test involves asking users to indicate what they would click on to complete whatever task you are monitoring. According to research carried out by Bob Bailey and Cari Wolfson, if the first click is correct, users have an 87% chance of completing the action correctly, as opposed to just 46% if the first click was wrong.
This kind of test is ideal if you have altered the functionality of a page in some way or changed your calls to action.
Five Second Test
The final test is called a five-second test. As the name implies, you show the user a page for five seconds and ask them what they recalled seeing.
That can be a useful test if you have changed the visual hierarchy of the page or made attempts to draw attention to certain screen elements such as calls to action. It will enable you to see if people can see what you want them to see.
Fortunately, running all three of these tests is easy using a tool called Usability Hub. They not only allow you to run all of these tests, but they will also recruit participants for you.
By carrying out these tests, you will be much more confident that the solution you have developed is correct and will avoid wasting money building something that might ultimately not work.
Once you are confident, you can go ahead and roll out the new update to the live site. However, that doesn’t mean you have finished.
Repeat the Cycle
The process of optimising a website for conversion is ongoing. Once you make one update, it is time to move on to the next by repeating the above process.
You return to the analysis stage, seeking out the next underperforming page and attempting to understand why, before starting to explore solutions.
In this way, you can ensure you have a robust working methodology to improve the effectiveness of your site over time.
