26th October 2012
Landing Page Optimisation: A Guide to Landing Page Tests
The results of an online marketing campaign are greatly influenced by the quality, design and content of landing pages.
A great set of carefully chosen PPC keywords and well optimised ad text will only get you so far. If your website landing pages do not meet your visitors’ expectations for one reason or another, your campaign will most certainly suffer.
Even if you are confident your current landing pages are performing well, optimising them by performing landing page tests could see them do even better.
This blog post will suggest some landing page tests you can do, how long to run the tests for, how to know if they’ve worked and how to build on the test findings.
Before You Start
As is the case with any marketing effort, offline or online, you need to define what you hope to achieve from your landing page tests. Your goal might be to increase the landing page conversion rate or perhaps your goal is to reduce the page bounce rate.
Types of Tests
There are typically two types of landing pages tests you can do. With A/B tests an alternative version of a landing page is created. This alternative/new page is showed to a certain percentage of the page visitors. So for example, you might show the original landing page to 50% of the URL visitors, the remaining 50% would see the new version of the landing page. If you wanted an experiment to be less risky, you could show the new/alternative landing page to only 20% of the URL visitors.
With multivariate tests different versions of the landing page are not shown, instead these experiments test different elements on the page. So for example, your visitors will all visit the same page URL but 50% of them might see an orange call to action button, whereas the other 50% might see a green call to action button. And 50% of visitors might see the button (of whatever colour) in position A, and the other 50% in position B. In such a test, two variables (button colour and location) are being tested simultaneously.
Choosing and Using a Testing System
Don’t Do Too Much At Once
If you are running A/B style tests it’s a good idea to change only one item of a page on each page variation. So for example, you may be running an A/B test where you have your original page and then two variations of it. On the first alternative page variation you might just change the page heading and nothing else. On the second alternative page you might just change the position of your call to action button and nothing else. If either of the alternative page variations performs well you will be able to easily assign the page change to the performance change. If you change too much on your new page design and the page performance completely drops off you won’t know what the specific cause of the drop off was.
What to Test?
There are a huge number of things you could test on a landing page. This list gives you some basic ideas.
- Call To Action
- The shape of the button
- The colour of the button
- The size of the button
- The call to action text e.g. “Buy Now” versus “Proceed to Checkout”
- Where the call to action resides on the page, you could test placing it higher up on versus lower down on the page
- Page Layout
- Move items on the current page around, for example, give images more prominence or make videos less prominent
- Try shorter or longer landing pages. Depending on your industry/product/service there might be a need for more or less information on a page
- Conversion Process
- Shorter forms
- One page versus two page conversion forms
- Page Content
- Headings: Test different text, fonts, colours, sizes and positioning
- Images: Test larger images, smaller images and test these with less or more prominence. Test images that do or do not contain people, add industry logos if applicable and test giving these more prominence
- Videos: Try adding videos to a page, test video prominence or try testing user videos versus stock videos
- Page body copy: Try increasing the amount of copy, test laying out text in a bullet pointed form, test the page text - for example try different introductions or talk about features and benefits rather than just product facts
The tests above can be implemented without any site rebranding; this is why we’ve not suggested changing complete colour schemes. This is something that needs some careful consideration.
How Long Should You Run a Test For?
This depends on the amount of traffic your page/test receives. If a test was implemented on a page that receives tens of thousands of visits a day, after a day you may know if a test has worked or not. You need a significant amount of traffic to truly know whether a test has worked.
However, remember to consider any external influencing factors and how these might impact upon your landing page tests. For example, if you know on Mondays more people buy from your website than they do on other days, it would not be a good to idea to run the test on a Monday alone, no matter how many visits the page gets.
Also remember to make a note if you have any special offers alongside your landing page tests. Offers, like discounts, will affect conversion rates without any landing page changes.
If you have a long conversion pipeline then this should be factored into your decision making too.
Results can be measured by not only using the landing page testing system but by also using any website analytics packages you have. You can see if the alternative URLs lead to different user behaviours like a longer time spent on the website.
Rinse and repeat! Once you’ve found out what has worked implement it on your site and monitor it. Remember to make a note of what changes you implement (and when), this will help you when you look back at your stats. You can then go on to test further page changes alongside the new implementations to make for even better landing pages!
If the landing page tests appear to have made no difference to the performance stats - don’t give up! Consider testing more radical landing page changes and keep going until you find something that works.