A public library's stakeholders want six departments featured in a rotating homepage hero carousel — why do carousels tend to test poorly, and what would you propose instead?
answer
- who ever sees slide four
- motion reads as advertising
- content moves while being read
- compromise between departments, not users
- one priority plus a visible list
basics
~20 sCarousels hide most of their content behind time, users rarely see or act on later slides, and rotation interrupts reading. Propose one prioritised feature plus a visible list or grid of the rest, and measure it.
solid answer
~50 sCarousels usually exist to settle an **internal** argument — six departments each want the homepage — rather than to serve users. In usability testing and analytics they commonly show the same pattern: most engagement goes to the **first slide**, later slides are rarely seen or acted on, many users read moving banners as advertising and look past them, and **rotation** changes the content under someone who is reading or about to tap. For keyboard and screen-reader users the moving content adds its own barriers. A stronger proposal is **one prioritised feature** chosen by an agreed rule (season, deadline, campaign), followed by a **visible grid or list** of the other departments' items, all readable without waiting. If a carousel stays, make it user-controlled, with no autoplay or a clear stop control, and measure engagement per slide to test the decision.
go deeper
Recall that most users only see a carousel's first slide and that auto-rotation needs a stop control.
Explain the reasons carousels test poorly: unseen later slides, ad-like motion, content moving while read, anonymous dots and accessibility cost.
Propose and defend an alternative with evidence: per-slide data, a comparison test, and a prioritisation rule stakeholders can accept.
Treat the carousel as a governance symptom: settle homepage ownership with a scheduling rule, and decide whether the system should offer an auto-rotating component at all.
## Why carousels get built A **carousel** shows one of several slides at a time and rotates between them, automatically or on request. On organisation homepages it is often the answer to a **governance problem** rather than a user need: six departments — children's services, events, e-books, local history, adult learning, branch news — each want the top of the homepage, and a carousel lets everyone have it. That is exactly why it tends to disappoint: it optimises for fairness between stakeholders, not for what users see. ## Why they tend to test poorly Usability studies and analytics on homepage carousels have reported the same pattern for years. Treat it as a well-documented tendency to verify on your own product, not a law. - **Later slides go unseen.** Most users look at the first slide, if any, and leave or scroll before later slides appear; engagement is typically concentrated on slide one. - **Motion reads as advertising.** Large, moving banners resemble ads, and many users have learned to ignore that part of the page. - **Content moves while being read.** A slide advances just as someone finishes its headline, or changes under a finger about to tap, sending them to the wrong place. - **Low information scent.** Dot indicators say more slides exist but not what they hold, so users cannot decide whether any is worth waiting for. - **Accessibility cost.** Auto-rotation needs a stop control, focus and hover pausing, named slides and properly hidden off-screen slides; screen-reader users can otherwise hear content from a new slide with no sign the context changed. - **Performance.** Several large images load for content few users see. ## What to propose instead | Option | What users get | When it fits | |---|---|---| | **One hero plus a visible grid** | A single prioritised feature, then every other item readable at once | Most homepages with competing departments | | **A static list of featured items** | Scannable headlines, no waiting | Text-heavy news and events | | **User-controlled carousel, no autoplay** | Browsing on request, nothing moves by itself | Genuinely sequential or visual content, such as new-arrival covers | | **Personalised or rule-based hero** | The most relevant item for this visit | When there is data to choose from and a rule stakeholders accept | The key move is to replace 'everyone gets a slide' with a **prioritisation rule** that stakeholders sign off once: time-bound campaigns first, then deadlines, then a fixed rota for the hero slot, with everything else visible below. ## Making the case to stakeholders 1. **Show their own data**: impressions and actions per slide. Departments on slides four to six usually discover they are getting little. 2. **Run a comparison**: carousel versus hero plus grid, measured on clicks through to each department's content and on task success in a short usability test. 3. **Offer a fair rule**: a rota or campaign calendar for the single hero slot, so the compromise moves from rotation to scheduling. 4. **Keep the carousel's one real strength**: if a sequence of images is itself the content — covers of new arrivals — a user-controlled carousel below the fold is reasonable. ## If the carousel stays - No autoplay, or autoplay with a visible stop control first in focus order, stopping on focus and hover. - Slide names or visible titles on the picker controls, not anonymous dots. - Slides that are fully hidden when not shown, and a slide count such as '2 of 6'. - Per-slide measurement from day one, so the decision can be revisited with evidence. ## What to watch for in the numbers Per-slide data is easy to misread. A few checks keep the argument honest: - **Separate impressions from views.** A slide that was rendered but rotated away after two seconds was not really seen; count only slides that were on screen long enough to read. - **Compare actions, not only clicks.** A click that lands on the wrong slide because it changed under the finger inflates the numbers of whichever slide came next. - **Look per department.** An overall rise in homepage clicks can hide the fact that three departments now get nothing. - **Include assistive-technology users** in any usability sessions, since the carousel's costs fall hardest on them.
- How would you test whether the replacement actually performs better than the carousel?Define the outcome first — visits to each department's content from the homepage, and success on a few realistic tasks such as finding this week's author talk. Then compare the carousel with the hero-plus-grid design, either as a controlled experiment on live traffic or in a moderated usability test with representative library users, and look at per-department numbers, not only the total.
- Is there a case where a carousel is the right component?Yes, when the items are a browsable sequence of similar things and the user chooses to move through them — a strip of new-arrival book covers, or product photos of one item. The user is in control, nothing moves on its own, and seeing only a few at a time is acceptable because the rest are the same kind of thing.
saying these in an interview costs you the question
- A carousel shows more content, so more users see more messages
- Faster rotation fixes low engagement with later slides
- Dot indicators tell users what the other slides contain
- Giving each department a slide is fair to users as well
- Auto-rotation is harmless because users can wait for their slide