Searchme Coverflow
Coverflow, like any other popular design pattern, is widely used and abused. It’s bling for your iPod and MP3 collection. It’s makes file browsing all fun and games. But please, keep it away from my search results. Just because it looks good doesn’t mean it works good. I seriously doubt that search result coverflows make it easier to find what I’m looking for. If you’re a company and you’re bent on innovation, don’t make the mistake believing that a different user experience is necessarily better. It’s more likely to be totally useless. Here’s why I think you should avoid search result coverflows.
Apple introduced coverflow as a highly visual and playful way of browsing music collections and computer files. The touch screen technology in iPhone and iPod Touch further amplifies the interactive aspect. It feels close to flipping actual physical covers. The popularity of coverflow is a testament to it’s usefulness, and it’s ability to make something as dull as file browsing at least a little more engaging.
Searchme were among the first search engines to present search results as coverflows. Web page thumbnails are displayed as covers, and you flip your way through the results starting with the highest ranking hit. The interface is visually appealing, even though the Flash animation is a bit sluggish. The thumbnails are supposed to help you determine how relevant each page is to your query. Clicking on a thumbnail takes you to the target page. Searchme looks good, but it’s not a good search user experience.
Thumbnail images don’t work for text. They work well for photos and graphics, but text does not provide enough visual cues to differentiate between miniatures. Web pages are still mostly text, menus and illustrations aside, and the text is vital when assessing the potential relevancy of a search result hit. Searchme is aware of this problem, and provides an optional text summary for each thumbnail, much like you would expect from any search engine. The fact that this option eventually became the default setting goes to prove that thumbnails alone are not a sufficient visualization of web search results.
Coverflow search results are ordered from left to right, deviating from established conventions. We have been trained to read search results from high to low, sorted descending on relevancy. It’s a spatial metaphor we are used to, but there is really nothing spatial about high and low relevancy. There’s no natural order in the real world placing more relevant things on top of other, less relevant things (although Monty Python would have it otherwise). Same way that placing things left or right of each other (like in a coverflow) makes no difference, either. But does it mean that coverflow should become a convention alongside the old, familiar list of search results? No.
The high/low spatial metaphor is very strong, and we use it all the time. We talk about tones having high and low pitch. We feel in high spirit, or we try to lower expectations. Left/right is undoubtedly a weaker metaphor. Since Arabs read from right to left, will they need reverse coverflows? How about Manga fans? And the left-handed? Perhaps they have a different sense of left/right relevancy.
Coverflow is a design pattern for browsing collections, not ordered lists. What do I mean by that? Items in a collection may certainly be ordered (sorted on title, date, file type etc), but there’s no implicit more than / less than relation between the items. A file is not more important simply because its name begins with the letter A. But a list of search results have such an order. The first hit is more relevant than the next one, and the one after that, and so on. Coverflows do not convey that difference in importance as efficiently as lists ordered from high to low.
And that is why coverflow makes no sense for search.
Search results are not just about the content of the hit, if your result list contains hits from may different types of sites, searchme has one advantage. It is possible to spot the type of page without even reading the text. I use searchme for one thin only. Finding product reviews. When searching for product reviews on Google I usually end up with 50 – 80% of the hits coming from price comparison sites with no review of the product at all, just a “review this product” link.
Using searchme, it is possible to quickly skip those pages in the results and scan the summary of the pages that actually looks like a product review.
The problem with Searchme is that they have put all their UI focus on the coverflow part, and very little on the surroundings. The textual result list below the coverflow view has barely enough information to distinguish the various pages. Both the text version below, and the “textual summary” overlay on the coverflow appears more like Google text ads than anything related to the search results.
What really strikes me about Searchme is that the engineers are the ones doing all the interaction design. The site is stuffed with features, but is missing the really useful experience.
Andreas
@Andreas Ringdal
Good point about spotting page types with thumbnails. Visual cues are important.
I really like the GooglePreview addon for Firefox (http://ackroyd.de/googlepreview/). It inserts page thumbnails into your regular Google search result. Very useful for the reasons you explain. I hope Google integrates it into the standard UI.
There’s a tabloid quality to the title of this article, slightly provocative, screaming for comments! Although, I agree with your conclusions regarding general search results not being fit for cover flow (or is it the other way around). However, there is plenty of synergies between this visualization and search.
In a ranked perspective the left-to-right browsing becomes confusing, but consider search results ordered in other ways (eg: chronologically).
Consider site-search, where the images are more structured (following a format): employee pictures, car models, snowboard designs.
-Thomas
@Thomas Kjelsrud
I think coverflow was initially conceived for browsing collections ordered chronologically or alphabetically. That means it should work well for search results if they are sorted this way, aswell. But web search results (like Searchme) are usually sorted by relevance, and that becomes a usability problem.
Both Cooloris/Piclens (http://www.cooliris.com/) and TagGalaxy (http://taggalaxy.de/) are good examples of well-functioning visual image search. It’s like you say, thumbnails work well for images, but they tell you little to nothing about the text content.
Do you know about other cool visual search engine (even coverflow ones
)?
I do not think it is the sort order that causes the problem. the problem with search results is that what you are looking for is not in the visual details, but in the textual content.
Coverflow shold be used as an additional tool for browsing search results, not as the main presentation.
Coverflow and piclens are both useless when searching wikipedia, no matter how you chooste to order the results.
andreas
Note this about Searchme: http://www.quantcast.com/searchme.com#traffic
Judging from the traffic increase, people do like it. I’m a very, very occasional searchme user.
@Otis Gospodnetic
Good point!
The stats say that 26% of the regular (returning) users generate more than 60% of the traffic. That’s not bad, I guess. How would you compare it to other search engines?
I find the demographics data interesting too: http://www.quantcast.com/searchme.com#demographics
It says Searchme’s users are (compared to the average Internet user) predominantly afro-american or hispanic males, teens or young adults, with children, low income and no college education. Not a very geeky crowd, I’d say.
I would love to dig deeper into stats like these. Maybe there’s hidden clues to why people like the Searchme coverflow experience.