Getting started with Conversion Rate Optimisation is easier than that you might think. As with a lot of digital marketing disciplines, CRO only requires that you follow a logical pattern of thought to get things started in the right way. Having said that, making sure you set up correctly and build on the right foundations will be key to getting the most out of your CRO activity. I’ve detailed a step-by-step guide to help you and your business become CRO ready.
Step 1. Identify your goal(s).
As with all things, if you don’t know what you’re trying to achieve, you’ll never get there. You’ve set up your website for a reason, so that will probably serve as your principal goal. But there may also be additional goals which help to feed that broader purpose. Write down your primary, secondary and tertiary (if appropriate) goals for your Conversion Rate Optimisation programme and use them as a reference as you consider different improvement ideas. The list will help you to prioritise where and when you need to align your resources most efficiently. Good examples of goals could be “Revenue Maximisation”, “Lead Generation”, or “Mobile Sales Volumes”. Goals should be broad enough to give you scope to work in, but distinct enough that you can judge performance against them.
Step 2. Reference your data.
With Google Analytics being freely available, there is no excuse not to have a reasonable standard of tracking set up for your website. So if you’ve reached this point and you aren’t using an analytics platform to monitor your site performance, please stop reading and get one set up! But assuming you do have site performance data available, now is the time to start putting it to good use. From your goals in Step 1, you may have identified ‘Sales Conversion’ as a key metric for your business, but you will only learn how to improve it by looking at your historical data. Don’t skimp on this step, as referencing your data will help you identify the best opportunities to achieve your goals. And they will be the best opportunities as they represent the biggest gap between where you currently are and where you want to be.
Step 3. Choose a test page.
By looking into your data in Step 2, you should have been able to identify areas of sub-par performance; now it’s time to narrow those areas down to a single page* you want to improve. A couple of things to consider when selecting that page: What sort of traffic and conversion volumes does that page have? There’s little point trying to run a test to a handful of visitors in a month, and if that page has little bearing on your key goal(s), is it likely to have the impact you are looking for? There’s no “Golden Rule” on volumes, but generally speaking, if you choose pages with 100–125 conversions per experiment per month, you stand a good chance of getting a reliable result. Where does that page appear in your goal flow? A basic rule of thumb is that the closer a page is to the visitors’ end goal, the greater the likelihood that you will be able to test with a significant improvement. This is of course because the (hopefully!) positive impact your test is having isn’t being diluted by too many steps afterwards. Every goal flow has its limiting factor and if visitors are hitting that after seeing your test, it can dampen an initially promising result. It is, of course, possible to run tests over multiple pages at the same time, but as this is about getting started with CRO, my firm recommendation is that you start simple and grow the complexity as you go.
Step 4. Generate a hypothesis.
Creating a hypothesis is a hugely important step in the process. As with setting goals at Step 1, your hypothesis sets out what impact you believe your change will have and it is essentially this belief that you are testing against, looking to prove or disprove it. And in order to do so effectively, your hypothesis must be specific. A good example of a hypothesis would be:
By adding an additional “Buy Now” button which can be seen when the page loads on a mobile screen, mobile visitors will find it easier to add products to their Basket, and therefore we will increase overall Add to Basket conversion.
What makes this a good hypothesis?
- It describes the change being made (the addition of the button).
- It describes the direct impact of that change (mobile visitors will find a button more easily)
- It describes the overall business impact (therefore ‘Add to Basket’ conversion will increase).
Step 5. Select a testing tool.
You will need some software to help you to deliver an alternative experience to your visitors. There are a plethora of tools in the market today, with the majority providing very similar functionality. Initially, it may sound like a tough call to make on which you should go with, but many offer 30-day free trial periods, so you can try out a few and find the one you like most with no obligation. Also, it helps to ask around too — Conversion Rate Optimisation is an ever-growing industry, and chances are someone you know is already doing it and might be able to give you a good recommendation.
Step 6. Build your test.
Now that you’ve chosen your optimisation tool, it’s time to build your test. Virtually every optimisation tool in the market today will allow you to do this without requiring web-language knowledge, although it does still come in useful at times! When building your test, try not to get overly hung up on small details. Yes, it is important that the test looks natural on your site, but remember that it is only a test — so if it does prove to be a winner and you need to neaten up the design a little before you deploy it permanently, then it’s no big deal.
Step7. Check your test as rigorously as humanly possible!
Once your test is built, try to check that it looks right in as many different browser & device combinations as you can. And if possible, try to get someone else to check what you’ve done as well. This may sound obvious, but even enterprise-level businesses have knocked their websites out of commission by not checking their tests thoroughly before launching them. You’ve made it this far, now is not the time to start cutting corners!
Step 8. Launch your test and monitor.
Once your test is live, do your best to monitor its performance. The first thing you’ll want to check is that you have data coming through for all of the metrics you want to track — without this, you’ll have no idea if it’s helping or hindering. Monitoring is necessary, but try not to fall foul of a knee-jerk reaction. Often tests can start off poorly as regular visitors may be a little thrown off by something new, but over the course of a few weeks, results can turn around. Likewise with tests that start strongly, don’t be too keen to declare a winner — try to give your tests an absolute minimum of two full weeks (14 days) to run before you look to pass any judgement.
Step 9. Check your results.
If you think your test has reached a conclusion, how do you check? Have you used a statistical significance calculator (very clear one here) to confirm the validity of your results? If you were trying to improve your order value, you’d need a T-Test check for average value changes (again, good one here) to be certain. And once you have your results, how do they compare to your original hypothesis? Did you manage to prove what you thought was correct, or has your hypothesis been disproven? Regardless of the result, it’s important to form a new hypothesis from your results for your next test, as this is what keeps your programme moving forwards.
Step 10. Record your results.
You might think that declaring a result is the final stage, but documenting your results should not be undervalued. Lay out what you tested, why you tested it and what your results show — this will be crucial in learning from all of your tests and you have somewhere to reference those results in future. Exactly how and where you choose to keep your results depends on who is involved in your Conversion Rate Optimisation programme and where they are located, but there are some basics to follow. Firstly, you’ll want to have a formal report of results; PowerPoint is usually a good medium for this, especially if you have a strong test and the CMO wants to know what happened! Secondly, and particularly important if you run a lot of tests, you’ll probably want a more straightforward tabular format such as Excel to show the basics — test numbers, key metrics, uplifts, etc. Finally, you may want to store them in the cloud somewhere (Google Docs, Trello, etc.) so they’re easily accessible. If you do store them somewhere like that, be sure that you use strong passwords to prevent access to sensitive data, and that any information security teams you might have are happy for you to do so.
Best of luck with your first test!