Should Marketing Sites Still Use Carousels?

Should Marketing Sites Still Use Carousels?

Almost never in a hero, and sometimes lower down the page. A homepage carousel is usually a way to avoid choosing what matters most, and the research says people mostly do not see past the first frame anyway. There are real exceptions, and they are narrower than the requests we get.

This comes up in nearly every project. Several teams each want their message above the fold, and a slider looks like a way to give everyone what they asked for. It is not a design decision at that point. It is a truce.

So here is what the evidence says, what accessibility rules require, and when we would actually build one.

Why Do Carousels Keep Getting Built Anyway?

Because they solve an organisational problem rather than a user problem. Product wants the new feature, sales wants the enterprise message, marketing wants the campaign. A carousel promises all three without anyone losing an argument, and it is very hard to say no to that in a meeting.

There is a second reason that is more sympathetic. A carousel looks like more content in the same space, which feels efficient when you are staring at a wireframe. The cost only shows up later, in behaviour you never observe because nobody reports the slides they did not see.

Nielsen Norman Group names this directly. It warns that carousels give designers a false sense of security, because the team assumes people will view all the content when in practice many will not.

What Does the Research Actually Say?

That most people never get past the first frame. NN/g's write-up on designing effective carousels reports that users often immediately scroll past these large images and miss content beyond the first one. The slides exist. The views do not.

There is a comprehension cost too, not just an attention one. NN/g notes that users see only one image at a time and may form an incorrect impression of an organisation from that single frame. If slide three is the one that explains what you sell, plenty of visitors leave with the wrong idea of what you do.

Its eyetracking work also gives a blunt figure on animation more generally: animated ads get looked at 27 percent of the time. Movement does not reliably buy attention, and on a homepage it often reads as promotional noise rather than content.

Do Carousels Break Accessibility Rules?

Auto-advancing ones usually do, unless you add controls. WCAG success criterion 2.2.2, Pause, Stop, Hide, is a Level A requirement, which is the baseline level rather than an advanced one. Failing it is not a nice-to-have miss, it is a baseline failure.

The normative wording is specific. For any moving, blinking or scrolling information that starts automatically, lasts more than five seconds, and is presented in parallel with other content, there must be a mechanism for the user to pause, stop, or hide it.

Auto-updating content is held to a stricter standard again. W3C's understanding document explains that the five second exemption does not apply to auto-updating information: users must be able to pause, stop or hide it, or control the frequency of the update, unless the auto-updating is essential to the activity. An auto-rotating hero falls squarely inside that.

How Long Should a Slide Stay Up?

Long enough to read comfortably, which is longer than most carousels allow. The five second figure in WCAG is not a recommended slide duration. W3C explains it was chosen because it is long enough to get a user's attention, but not so long that a user cannot wait out the distraction if necessary to use the page.

That framing is worth sitting with. The standard treats automatic movement as a distraction to be endured, not as a feature to be tuned. If your slide timing is a design debate, the underlying question is whether the movement should exist at all.

And whatever you pick, respect the reduced motion setting. A visitor who has asked their operating system for less animation should get a static first slide, not a slower carousel. Our notes on designing for reduced motion cover how to wire that up.

What Do Carousels Cost in Performance?

More than their file size suggests. A hero carousel usually loads several large images at once, and only one of them is visible. That means bandwidth spent on content most visitors will never see, competing directly with the image that decides your Largest Contentful Paint.

Then there is the library. Most carousels arrive as a JavaScript dependency with its own CSS, which adds main thread work at exactly the moment the page is trying to become interactive. On a mid-range phone that is the difference between a hero that feels instant and one that feels sticky.

The third cost is layout stability. A slider that initialises after paint can resize or reposition its container, which is the classic recipe for a layout shift in the most visible part of the page.

When Is a Carousel Genuinely the Right Answer?

When the items are genuinely equivalent, genuinely browsable, and the user came to compare them. A gallery of project images, a set of customer logos with quotes, a product photographed from several angles: in all three the visitor knows more exists and has a reason to look.

The distinguishing test is whether the user is in control. A carousel someone drives with intent is a browsing interface. A carousel that moves on a timer while someone is reading something else is an interruption. The same component, two completely different experiences.

Position matters too. Lower down the page, where visitors have already understood what you do, a slider costs far less. The damage is concentrated in the hero, because that is where your one clear message is supposed to live. Our piece on hero section design covers what belongs there instead.

How Do You Build One That Does Not Fail People?

Five rules, and they are not optional. Do not auto-advance, or if you must, provide a visible pause control that meets 2.2.2. Show clear position indicators so people know where they are and how much is left. Keep it to about five frames, which is where NN/g says recognition of already-seen content starts to break down.

Make it keyboard operable, with real focus states and arrow key support, and make sure screen readers are told when the visible slide changes. A carousel that works only with a mouse or a swipe excludes people for no design benefit.

Finally, load only the first slide eagerly and defer the rest. NN/g also advises avoiding auto-forwarding on mobile specifically, because people scroll fast enough that slides change while they are already looking at something further down the page.

What Should You Use Instead on a Homepage?

One clear message, then the other messages as sections underneath. Scrolling is free, universally understood, and does not hide anything. The content you were going to put in slides two and three still exists, it just exists where people can find it.

If the audiences are genuinely different rather than the messages, a segmented entry point works better than a rotation. Two clear paths labelled by who they are for beats a slider that shows one audience their message and the other audience nothing.

And if the goal was to show range rather than to communicate a message, a static grid does that better than a carousel. A visitor sees six things at once instead of one thing six times. We covered the movement question more broadly in our piece on website animation and conversion.

What Would We Do With a Hero Carousel Request?

Ask what happens if only the first slide is ever seen. If the answer is that the page still works and still says the right thing, the carousel is decoration and can go. If the answer is that something important disappears, that something needs its own place on the page rather than a rotation slot.

That question tends to end the debate faster than any research does, because it turns an aesthetic argument into a prioritisation one. Somebody still has to choose, but at least everyone is now choosing knowingly rather than hiding the decision inside a component.

If you are stuck in that conversation right now, we are happy to help you work through it. We do this with B2B teams regularly at phoenix.studio, and the answer is usually a clearer hero and two extra sections rather than a better slider.