We pride ourselves on driving our design decisions based on analytical data. Erik Runyon featured a post last January on his blog outlining, with empirical data, the relative ineffectiveness of carousels. His test source was the University of Notre Dame homepage and his conclusion created a mandate to our industry.
The Case Against Carousels
Runyon ran a test using Google Analytics for a 3-month period on a carousel prominently featured on ND.edu. He found only 1% of nearly 29,000 total clicks was on a carousel item. Of that 1% of clicks, 84% were on the item in position 1 with the rest split evenly between items 2 – 5.
He continued the test on other Notre Dame properties which had inherently more specific audiences as opposed to the homepage which had completely divergent user interests. The number of clicks on the carousel was between 1.7 and 2.3% – somewhat higher and to be expected given the nature of the audience. The same trend from the home page continued: item 1 garnered 48 – 62% of clicks with subsequent items splitting the rest evenly.
The stats improved slightly for overall clicks on secondary positions for auto-playing carousels. But auto-playing carousels carry some warnings along with it. Jakob Nielsen ran similar tests with a very similar conclusion. His conclusion is that if a carousel is to be used, then it should only rotate at the user’s direction. We agree.
Generally speaking, users want to be in control of the pace of content consumption and nothing is more annoying than reading part of a carousel only to have it auto rotate and leave you searching for a means to get back – if you even attempt to go back in the first place. Auto-playing carousels adversely effect low-literacy users, international users, and others with accessibility issues as well.
You can get a bit smart with an auto-playing carousel implementation and have the slideshow pause when the user has his or her cursor over the element – but that brings up mobile concerns of its own since the concept of a hover is completely nonexistent in a mouseless world.
It’s easy to see that Runyon and Nielsen’s data draw a very simple conclusion: don’t use carousels. And that piece of advice should be heeded as a general rule.
The Case For Carousels
Carousels aren’t without merit. It’s important to explore the nature of the carousels in the experiment. For the ND.edu homepage, the carousel featured news stories that may not be pertinent to a user’s interest in general. Low click-through rates are to be expected if the content of the carousel isn’t contextually sensitive with the information on the rest of the page. It would be interesting to see the data on carousels that operate as a part of the fabric of the rest of the page’s content. For example, a firm may choose to employ a carousel on their services page to allow the user to navigate the content in a more meaningful way.
A case could be made that the purpose of some carousels is meant more to influence a particular emotion vis-a-vis design rather than being 100% informative. This approach could be particularly useful for non-text media. Image and video galleries may enjoy some upside to using this format where it falls completely flat with traditional text-based media.
Carousels simply display a lot of information in a condensed space and toggle that content either automatically or based on the user’s direction. So what would be a good alternative if we were to magically destroy carousels overnight? Instead of condensing the vertical space used to portray that content, we’re finding a lot of new design trends are opting to display that same information by tiling content panels vertically. The same user interaction exists but the hurdles to the operation are much lower: the user simply scrolls down.
Scrolling is a very passive action. In fact, you may find yourself scrolling down before reading anything at all on a particular page. Most every user is going to give you a scroll. That number is drastically lower for those giving you a click. Displaying the same content vertically as opposed to a fixed height will allow the user to consume the same information but completely on their terms. It also makes for a nice and elegant magazine style layout and it leaves room for a lot of creativity in between.
We agree with Runyon and Nielsen. Carousels are, by their very nature, ineffective. The proof is in the data and we listen more to data than we do emotion. We find ourselves defending this conclusion with our clientele who will often ask explicitly for a carousel. After empowering our clients with data, they often will acquiesce in the presence of a better alternative. We never shoot down ideas but do provide context for our advice.
Runyon and Nielsen have challenged web designers, developers, and content strategists to get more creative in regards to showcasing content that would have traditionally been given the carousel treatment. Our advice is to listen to the data. If carousels must be used, be sure to use them within the context of the page’s surrounding content. Use them to allow users to navigate rich media as opposed to giving them small, often meaningless, thumbnails. Finally, don’t use them if position 1 is as important as position 5. If that is the case, challenge your team to get creative in terms of design and content strategy.