Several months ago, we made some changes to the search results of hotels.com, and among these was the creation of a “pinned header”. As you scroll down through the list of results, a portion of the page header stays with you. Here’s the UI before scrolling. And here it is with the header pinning (linking to screenshots for archival purposes).
The rationale for this change was that the business wanted us to shout as loudly as possible about our loyalty programme called Welcomerewards. But I was worried that if we did this in a conventional way (eg a big banner at the top of the page, or worse, dispersed inside the hotel listing) it would simply fall victim to “banner blindness” and be ignored.
So in an effort to allow the loyalty scheme message to be noticed without relying too much on visual design to achieve that, and also to improve the prominence of the sort control (which we were also trying to encourage more use of), I thought we should try pinning it to the top of the page. My hypothesis was that customers wouldn’t notice the message (or the sort control) at first, but once they started scrolling, it would pop out at them without being intrusive. I based this idea somewhat shakily on the fact that humans and other animals are good at noticing movement in peripheral vision.
I was disappointed, therefore, to discover during user research at the end of last year (lab tests in London, Rome, Dallas, Sao Paulo and Shanghai) that nobody appeared to notice it. We didn’t do any eye-tracking in that study, but I was hoping that at least one or two people might either remark on the fact that header allowed them to sort at any time, or perhaps inquire about the loyalty scheme. But not only was it not noticed by the test participants, even the note-taker himself didn’t see the pinning until three weeks into the testing when I mentioned the issue!
The idea has therefore been a failure. But why?
The answer may be in some recently-published research that shows that humans are surprisingly bad at noticing when moving objects change. Of course, it’s less than straightforward to make the connection between this research and what I’ve observed with the pinned header, but I wonder if it is related?