How to get started with conversion rate optimisation (CRO)

Optimisation/Conversion

So, you’ve read and hopefully now understand the basics of CRO. You’re probably thinking, “Where do I start with it all?” Well, have no fear because we’re sharing with you a quick guide on how to get started with CRO, as well as any top tips we’ve learned along the way.

Some businesses still feel that conversion rate optimisation isn’t important - which to us seems a little archaic. If your business is one of these, then just think, on average 98% of your total visitors to your website will never return - that’s a huge percentage! Marketers spend so much time working on getting users to site but once they’ve reached that goal, it seems they think the work is done. It’s time to stop wasting opportunities and start converting these users to give them a reason to come back. So, time to dive into what we think are the most important things you need to know.

1. Start by auditing your website

One of the first things you should do before even looking at the different optimisation tools (Yes, we know there are a lot and it can be confusing, but we’ll get to that later) is to identify the pain points on your site. There are a few ways of doing this but our number one tool is Google Analytics, closely followed by heat mapping software. If your site gets quite a lot of traffic then we suggest running survey tools on your site, such as Hotjar, before ideation and planning of tests. If you’re still a small to medium sized company, I would stick to analytics and heat mapping software initially. That’s not to say you shouldn’t look at surveys but, if you’re not getting a lot of traffic to your site, you may be disappointed by the lack of responses. You could be wasting time waiting for people to answer the survey when you could have started testing.

If you don’t get much traffic to your site currently, then I would recommend implementing the survey software once you’ve done a couple of tests and the users to site has increased. That’s when you should get the optimum outcome for these practices.

Google Analytics

As marketers, it may seem that we spend our whole lives looking through analytics and finding stats for the company’s monthly reports. However, this time actually look at some different KPIs, the ones that aren’t as pretty and that you may hide or not report on to the wider team. These could be things such as the average time spent on-page, the least popular pages on site or total pageviews. The most important stats to look at for CRO are bounce rate and exit rate.

Google Analytics should give you insight into which pages are the poor performing ones. These could be pages that have a really high bounce rate or pages that only a small number of users are visiting. Either way, these are the pages that you should start to look at for testing.

Heatmaps

You know which pages need improvements but now don’t know what to change and test on the page. If you want to gain a better understanding of how users are interacting on your site, this is where you could use a heat mapping software tool. There are various to choose from and it all comes down to personal preference (and budget). We love Hotjar’s version. Once set up, you can simply search for the pages that you’ve identified as needing improvements and the heatmap will track the user’s interactions on the page including their mouse movements, where they’re clicking and what they’re clicking, and how far they’re scrolling. All these actions are captured on a heat map which is laid over the page to provide a visual of where users most and least engage.

conversion rate optimisation heatmap

Survey tools

If you’re looking to implement a survey on your site or already have one set up then this can also be used to collate your audit and secure what actions need to be taken on the pain points you’ve uncovered. What better way to find out what to test than by having the users tell you what they want to see. We love Hotjar’s survey tool which includes a pop-up functionality of questions to ask on your site or page. You can change these questions yourself if you want to find out something in particular, as well as include multiple questions you would like to ask. Just remember that the user might get bored of answering if the questions keep on coming. We like to ask one question initially which is “did you find what you were looking for?” This is such a simple question but it uncovers valuable insights that can help when planning your tests.

Please remember, if you've received a lot of different feedback from a survey, that you don’t have to implement all these changes at once. Prioritise issues that are the most important and will ultimately help your business goals and KPIs.

2. Understanding how to test our site

After conducting your audit, you should know where and what the problems are. This is where conversion rate optimisation comes in to play as it’s all about making intelligent changes guided by date and real user experiences. Gut feelings can sometimes guide the right solution but by testing these out you’re letting the user choose which solution they like best. The two common types of testing are A/B testing and multivariate testing.

A/B testing

A/B testing can also be called split testing. This type of testing tests one variable at a time. Ideally you have the original page which is usually called the control. This is tested against the same page but with something altered such as a CTA button that's moved higher up the page or the colour of the CTA button being different. Whatever changes you make, this one is then usually called the variation.

Multivariate testing

Multivariate testing is very similar to A/B testing. The only difference is that it tests more than one variable at a time. Sort of like conducting multiple A/B tests but all at the same time.

We suggest A/B testing so you know which change has had an effect, be that positive or negative. Using A/B testing could seem a slower process but you're getting the best outcome for your users and business by knowing what’s succeeding and what isn’t.

If you’re a small to medium sized business or don’t have that many users coming to site, try A/B testing rather than multivariate. Your results might not be significant enough to be able to conclude which test has been successful.

Testing which changes to make to your site that result in a better conversion rate is always going to be crucial to CRO. However, don’t forget to use common sense. If you’ve found a general site improvement that would reduce page load times, then this wouldn’t need to be part of a test as all users would benefit from this regardless.

3. Start to implement these tests on your website

Sorry to disappoint but when it comes to optimising a website, there’s no right way. Different people like to use different testing tools. Here are some of the most popular ones:

The main thing at this stage is to align your marketing team with your development team (in-house or external). Your dev team might have some ideas they’ve thought of as well - after all, they’re looking at the website more than yourself. Brainstorm the ideas you’ve thought of and any they have and plan in time for the dev team to create these tests by using one of the tools above or others they like the look of. Remember to start with the most important changes first.

4. Knowing how long to run a test for and the best way of analysing

This part is where the size of your company and/or users to site impacts how long you should run a test for. The standard time that should be used is four weeks. This should give enough time for users to interact with the test as well as provide enough data to conclude the winner.

However, if it’s a large business and the test that you’re running has an impact on the whole company, the test may need to run for longer than four weeks. This ensures the results aren’t just because of seasonality or a special offer but because the variation is actually winning. After all, testing is still a risk to conversion so everyone needs to feel confident in the results.

Alternatively, if after a couple of weeks it’s obvious to see which test is winning, you may as well end the test and roll it out. There’s no point waiting a few more weeks when the goal has already been achieved as this gives you time to start implementing the next test.

How to analyse the tests

Depending on which software you use, there may already be a reporting tool that you can look at to gather data. But don’t worry if there isn’t, this is where we go back to Google Analytics. When your development team sets up the test you should also talk to them about analytics tracking and creating different segments for the versions of the test. This way you can search for that particular segment and know from there how it’s performing, by looking at the average time spent on the page, the bounce rate and the conversions, as well as adding the other segment and comparing them both.

5. Finally, don’t let it stop at one test

Ok, so the test you’ve been running for roughly four weeks has won, but don’t let the high of one test success make you feel like you’ve cracked it. There’s always other ways to improve your site’s performance and improve conversions. That test may have won, but do you think it’s worth building on that particular test again and making another improvement? Or is it now time to move on to another pain point that was identified in the audit? What we’re trying to say is never stop testing as it’s a great tool to use and one that gives your website a human factor. You're showing customers that you understand them and want to make their life as easy as possible.

On the flip side, please don’t feel down-heartened if the test has lost. It’s frustrating as you’ve waited weeks to realise that the control has still won. But always see this as a blessing in disguise. You’ve tested it and it hasn’t won, so you now know that whatever change you had made doesn’t work for the site and more importantly, the customers. After all, this is what testing is for. Be thankful you didn’t roll it out site wide and lose the momentum you originally had going - imagine that! Also, don’t think because that test hasn’t won that you should call it quits and move on to another page. Why not try an iteration of that test and run that one instead? It might be that some parts of the variation worked but not everything. You’ll never know until you test it again.