The color of the button has little to no effect on it’s own. What is more important is how it changes the visual hierarchy of the whole page, how it makes the call-to-action stand out. Plus additional information and wording of the button itself.
It’s also what we’re used to. Bing increased their revenue by $80 million dollars by finding the exact color of blue for their links. Why is that? It’s because people are used to links being blue.
Erik Runyon compiles some actual data on how people use carousels, including auto-forwarding carousels and those requiring users to click through. His conclusions:
First, if they’re going to insist on a carousel, they need to include compelling content that not only entices users to click, but can get their attention in the first place. Second, I might suggest keeping the number of features to a maximum of four (or better yet, three), as it appears that as the number of features increases, the click-throughs on sub-features decreases dramatically. Finally, I’d suggest that the subject matter can make a big difference.
Most website carousels are added not because of a focus on the consumer but because of internal organizational politics, Brad Frost points out. His first rule for when to add a carousel: “Make sure you actually need one.”
Meghan Keaney Anderson warns against personalizing content just because you can: “Personalization without a good reason breaks the cardinal inbound rule: Marketing should be driven by the prospective customer’s needs. At its core, marketing should be useful.”
Pilot programs and other trials can often lead us to the wrong conclusions, Megan McArdle warns:
This is one more installment in a continuing series, brought to you by the universe, entitled “promising pilot projects often don’t scale”. They don’t scale for corporations, and they don’t scale for government agencies. They don’t scale even when you put super smart people with expert credentials in charge of them. They don’t scale even when you make sure to provide ample budget resources. Rolling something out across an existing system is substantially different from even a well run test, and often, it simply doesn’t translate.