Almost all of the testing I've managed has proven that content delivered via carousels are missed by most users. Few interact with them and many comment that they look like adverts — we've witnessed the banner blindness concept in full effect.
In terms of space saving and content promotion, a lot of competing messages get delivered in a single position that can lead to focus being lost.
In all the testing I have done, home page carousels are completely ineffective.
For one, anything beyond the initial view has a huge decrease in visitor interaction. And two, the chances that the information being displayed in the carousel matches what the visitor is looking for is slim. So in that case the carousel becomes a very large banner that gets ignored. In test after test the first thing the visitor does when coming to a page with a large carousel is scroll right past it and start looking for triggers that will move them forward with their task.
The only exception I found was when testing around a holiday and the carousel spoke specifically to that holiday there was an increase in the amount of clicks a visitor had with the carousel.
We have built these for clients in the past with the main driving force being SEO. (Carousel images with text / links overlayed). They are a way to cram a lot of content onto the main homepage without looking like you are 'gaming' the search engines or keyword stuffing.
We do try to make them as efficient and usable as possible, but they are requested by the marketing people because they look 'modern' and provide the ability to increase the amount of copy on the homepage without bombarding the user with useless information.
They have also been used because different areas of the business don't want sites to give too much emphasis to 'X' product / service, so providing the master 'hero' image as a rotating image then various areas of the business can have their main product in the pride of place at the top of the homepage all at the same time.
So yes, in my experience they're primarily a marketing tool and not particulary built with the user in mind.
Carousels are effective at being able to tell people in marketing/senior management that their latest idea is now on the home page.
They are next to useless for users and often "skipped" because they look like advertisements. Hence they are a good technique for getting useless information on a home page (see first sentence of this post).
In summary, use them to put content that users will ignore on your home page. Or, if you prefer, don't use them. Ever.
By the way, these views are not my own, but are based upon observing thousands of tests with users.
I do not use or suggest the use of carousels. The changing of images can distract users when they read text on the page.
You might find some interesting information at http://digitaleskimo.tumblr.com/post/752912498/image-carousel-appropriateness
https://blinkux.com/ideas/usability-highlights-2008-beyond does not dispute the use or causeless, but offers some tips in the "Avoid giving users a confusing ride on your carousel" section.
As a user I find carousels faintly annoying:
Most have usability fail which I fall into the categories described in this article:
No ability to bookmark a particular item on the carousel, for example take a look at the BBC News photo carousel they use: http://www.bbc.co.uk/news/world-africa-14619799 There's no way for me to bookmark the sixth item and send a link.
Carousels that don't allow me to right click on an item and "open in new tab/page" - flash carousels
Carousels that have unpredictable non-intuitive navigation such as rotating content for no good reason just because you moved your mouse over it.
Having designed a lot of ecommerce websites I would say that carousels are effective. Specifically in getting users to view more main promotions. They are fun for users to interact with and improve the likeliness promos will actually be read or make an impact.
I think carousels can be effective as long as they give control to the user. That is, they can skip ahead, direct the flow, know where they are in the carousel, and turn off an auto-play function.
Here is more on this idea:
Approximately 1% of visitors click on a feature. There was a total of 28,928 clicks on features for this time period. The feature was manually "switched/rotated" a total of 315,665 times. Of these clicks, 84% were on stories in position 1 with the rest split fairly evenly between the other four (~4% each)... ("Feature" refers to the individual calls-to-action that are either manually or automatically rototated in and out of view.)
The article also discusses the difference between Static carousels (i.e. ones that require manual use to scroll) and Auto-forwarding Carousels. Surprisingly the Auto Forwarding ones recieve the highest usage (8.8% of visitors clicked on the carousel - 40% of those clicks were for the item on the first slide). However the article also references the Nielsen group article stating that auto-forwarding carousels are not a good user experience)
For Static carousels the average click rate was between 1.7 - 2.3%, again with the first item in the carousel recieving significantly higher selection (48-62%).
The main article source for this content is from weedygarden.net
I've also been struggling with the marketing team, that wanted on the one hand optimize the homepage load speed, and on the other - put a carousel with huge images "showcasing the amazing experience." And in this particular case it proved totally ineffective.
Here are some interesting finding on the topic: http://conversionxl.com/dont-use-automatic-image-sliders-or-carousels-ignore-the-fad/
In short - most conversion experts are against rotating banners and carousels and if you do need to take care on conversions, you'd better think of other options to "showcase experience" :)
I think the carousel has its use in e-commerce depending on the type (actually the size) of website that you deal with.
On the other side if you deal with a website where you have a very small range of products, there is no need for carousel (example: http://www.hardgraft.com/) and it is more than obvious that one of these tools would definitely disturb users.
A good exercise would be for someone to imagine getting to target.com for the first time and not having a carusel ... where would he go, what would be the bounce rate for the homepage etc.
Also there are examples of carousels on homepage on very small websites (http://cleaneverything.com/) but in my opinion on this example the carusel is useless.
From my own experience in looking at the analytics data of sites I've created, I can say that most users don't interact with a carousel, much less convert from one.
I have noticed recently that a number of sites that used to have carousels no longer have them and are instead showing just one "panel" (if you look at the HTML, there's still remnants of a previous carousel in some cases). Microsoft is probably the most noteworthy example (http://www.microsoft.com/) another is BYU's site (http://home.byu.edu/home/). Google analytic's site (http://www.google.com/analytics/) I think used to have one but no longer do.
While that doesn't directly answer your question (others have already done that well), I think its interesting to note that large organizations like Microsoft (who I'm pretty confident look at and analyze conversion data) have decided to ditch the carousel, probably in favor of faster load times.
Interesting sidenote: I think NNG's website (http://www.nngroup.com/) provides a good alternative for a site that wants to get rid of a carousel/hero image altogether. I've seen tons of sites with the exact same layout except for a carousel between the company description and the three blocks with images. Note how on NNG's site, you don't really miss having a carousel between those two page elements. In fact it's better without.
Most carousels have pagination arrows and dots. Users aren't drawn to this. They're drawn to text labels.
Labels are informative, meaningful and describe what users want. Labeling each slide incentivizes them to click because the labels tell them what they’ll get. Users are more likely to click on something that looks informative to them.
(Note: I hate the things, but this article seems to have come to peace with them and talks about how to use them the best way, if you must use them.)
Friends. Colleagues. Developers. A carousel can be highly effective.
Suppose your buyer is budget constrained and the bargains presented there are within their limit A carousel says "We're not sleeping" "We're still alive and kicking." "We're hep to the latest, greatest thing." "We've got great deals!" Otherwise, a carousel is a heavy data load, (Real estate vs. going out of your Intellectual property,) and hopefully has a balancing share of detractors.
I found out it's a big issue for accessibility as well. For example if the user only uses keyboard for input. If we are showing images it is pretty effective. Showing more data using carousal is not effective.