Page Performance Improvement = Conversion Rate Improvement
Page performance matters. Most of us think we know it. The SEO crew have been telling us this since Google confirmed page performance would impact search ranking back in 2010. But do we really design our pages with speed in mind?
Performance isn’t just an SEO thing. Many digital marketing disciplines benefit from a fast loading website. PPC, organic search, email marketing and of course CRO all benefit from well tuned pages. Apparently 40% of people abandon a website that takes more than 3 seconds to load. Whether it’s actually 2 seconds or 10 seconds doesn’t matter. The point is, the quicker your page loads the less “leakage” you’ll experience. It’s a lot easier keeping people on your site than trying to attract new users.
Above The Fold
Mobile Page Performance
We recently ran a page performance related AB test on a travel site. It must have been the easiest test to setup in the history of mankind. The control (live) version of the page was actually coded in such a way to slow down the animation of the search field. Visitors therefore had to wait for around 4 seconds after clicking in the search field before being able to start typing. Our test simply removed this animation and allowed customers to almost instantly click and start typing (less than a second). This change was on the homepage and yet it resulted in a significant impact on sales! We saw a massive +6% uplift on sales on mobile (and a positive result on desktop too, though not quite so high). All this from a simple code change leading to a fast loading page.
Don’t stop here though – continue to iterate on performance tests and you’ll see further benefits (we’re exploring this further with the site in question).
People Are Impatient
According to Google, people will visit a Web site less often if it is slower than a close competitor by more than 250 milliseconds (a millisecond is a thousandth of a second). Arvind Jain tells us that “Subconsciously, you don’t like to wait,” (Arvin is a Google engineer who is the company’s resident speed maestro). “Every millisecond matters.”
Back in the days of dial-up we had more patience. The human race has since evolved as we’ve benefited from faster and faster broadband speeds and now it’s become survival of the fittest (or quickest) websites. The same principles that make urgency messaging work also apply to page performance.
Can You “AB Test” Page Performance?
Interestingly, one of the downsides to AB testing is that there is a performance hit that comes with it. With the correct set up the impact on page performance can be minimal. Even so, page performance will always be worse when running an AB test as there is simply more going on in the background. Companies who are serious about CRO understand and accept this and simply go in with their eyes open. Ultimately it’s a small price to pay for the insight gained from testing. It’s a consideration nonetheless.
Split testing often impacts page performance the worst. The original page must load first when performing a client side redirect (which is how split tests are often implemented) in order for the redirect to then be initiated. People falling into experiment 2 (the new redirected URL) will therefore see a slower page performance than those in the control as they essentially have to wait for two pages to load rather than just the one. This is one reason that if you run a client side redirect test which is successful then your result is probably better than you think!
Despite the above you can of course test some elements of page performance. Just like we showed with the case study mentioned earlier (see mobile page performance section) it is possible to use AB testing tools to speed up certain interaction with elements on a page. About 3 years ago I ran a test where I hypothesised that changing the slow scroll on page links (to get to product reviews) to instant page jumps would result in a lift in sales conversion. It was a bit of a long shot but we tested it and got a strongly significantly positive result! People just don’t want to made to wait.
Avoid Carousels – They’re Not Worth It!
Carousels seem to be all the rage. In my experience they just aren’t worth it. They’re annoying to most users and slow down the page. Rather than loading one image you’re loading 3 (or more) causing a negative impact on page performance. You’re also wasting valuable space (usually above the fold) which could be better used to promote your products or services. I’ve run a number of different tests on homepage carousels but have never seen a result which indicated they’re worth having.
Perceived Page Performance
When running AB tests it’s not always about actual page performance being impacted – sometimes it’s simply that we need to wait for the original page to load a particular element so that we can transform it. Often an AB testing tool will hide the whole page until the element it needs has loaded. This increases the perceived page performance because if the test hadn’t been running then the visitor would have at least seen some elements loading in their browser even if others were still being processed. It is possible to only hide the element(s) that you are transforming and to let the rest of the page load in the usual fashion. This technique, known as a partial hide, is a much more elegant solution than employing a full page hide.
A Word On Tagging
If you don’t tend to run tests across most pages on your site then it may be worth reviewing your tagging strategy. When it comes to CRO you only need to tag pages that you want to test (i.e. transform) or track. Any other pages can be left alone and don’t need to suffer from any page performance impact. You don’t always need to wait for statistical significance either so once a test has completed you’re free to remove the tags. Just make sure you have an easy way to switch them back on when you want to build and run your next test.
In summary, page performance matters. Your conversion rate will suffer if your pages are slow to load. Even if they’re not really slow but just slower than your competitors you’ll still be impacted. CRO comes with a performance impact but it’s well worth it given the insights that can be gained from testing. Just make sure you’re being smart with the way you implement your tags and set up your tests. In my experience an experienced web developer is worth his/her weight in gold.
Author: Phil Williams
Phil is the founder of CRO Converts. He has had the opportuntity of creating successful testing and personalisation strategies for many of the UK and Europe’s leading brands.