Summary: Most users are unable to solve even halfway complicated problems with search. Better to redirect their efforts into more supportive user interfaces when possible.
Users are incredibly bad at finding and researching things on the web. A few years ago, I characterized users' research skills as "incompetent," and they’ve only gotten worse over time. "Pathetic" and "useless" are words that come to mind after this year's user testing.
In a recent study, for example, a user wanted to buy a highly protective yet girlish phone case as a gift for her daughter. While on Amazon.com, she engaged in random fishing expeditions into the product database, using search queries such as "pink impact resistant iphone 5 cover."
This was by no means the worst query we saw that week; in this case, however, the user never found what she wanted. She tried a few query modifications — which most users won’t attempt — but never questioned her basic research strategy. Nor did she realize that Amazon uses a full-text search that doesn't understand the meaning of a query.
Among the products she found were those that were not pink, not a phone case, and not an iPhone case. At least Amazon's thumbnails are good enough that she didn’t have to click through to most such erroneous hits.
Excerpts from first page of search results for "pink impact resistant iphone 5 cover" on Amazon.com
Amazon.com has poor usability in cases like this because of its business model: it carries every product under the sun. Such a design cannot support any one product category optimally, even though the overall site still scores better than almost anybody else on e-commerce design guidelines.
Easy Search Works; Difficult Search Fails
In study after study, we see the same thing: most users reach for search, but they don't know how to use it.
It would certainly be nice if schools would get better at teaching kids how to search. But I don't hold out much hope, because most people have the literary talents of an anteater (I was going to say, "a chimpanzee," but these animals are too smart for my metaphor). Having new and varied vocabulary words spring from their foreheads wasn't a survival skill for ice age hunters, so most people today can't think up good queries without help.
It seems a puzzle why ever-more users are so search dominant when they can't search. Perhaps because search actually does work as a user strategy in many cases — mainly when people use a web-wide search engine for simple tasks. Indeed, the big search engines (such as Google, Bing, Naver, Baidu, and Yandex) have morphed into answer engines that often give users what they want right on the SERP (search engine results page).
So, if you have a well-designed search facility, and users are looking for a specific item with a well-defined name, they'll probably be successful. In our testing of search on e-commerce sites, users found what they wanted in their first search attempt 64% of the time. And their overall success rate with search was 74%, which is pretty good.
The extra 10% success comes from users attempting multiple queries. Another way to look at this is that the 36% of users who failed their first search succeeded only 28% of the time (10/36=.28). In other words, the probability of success drops from 64% to 28% when we go from an easy to a difficult search problem.
(Intranets are a different matter: Because intranet search is usually very poorly implemented, most company employees have learned to limit their use of intranet search. In our recent intranet user testing, employees used search as their initial wayfinding strategy in only 19% of tasks on average—though with extreme variability between companies with good or bad intranet search.)
Designing for Mediocre Searchers
Search suggestions are a now a popular way to help people overcome their limited generative abilities by showing a drop-down of fully formed potential queries as soon as users type in a few characters that hint at their needs. Although helpful, search suggestions can also be limiting; users often view the drop-down as a mini-SERP and assume that it lists everything the site carries. Thus, if something isn't included in the search suggestions, users might never bother to search for it.
We found a better way to help users overcome their search problems when we tested people shopping at Costco's website. One user wanted to buy a TV set and immediately entered the query television on the Costco homepage. Following is the "search results page" that appeared:
Costco website as tested: the page shown after searching for "television."
Not a SERP at all! This is the category page for TV sets. The page includes relevant features for this product category, with relevant facets such as plasma vs. LCD, brands, resolution, and screen size. None of these would appear on a "normal" SERP, which has to handle a variety of result types.
Is Costco cheating? Only if you have a very strict interpretation of the search concept. In our test, users were happy on this site. They got what they asked for — which is exactly what people want.
For sure, you should redirect users from a normal SERP to a category page only when their query is unambiguous and exactly matches the category. A search for "3D TV" could go to the subcategory page for these products, but a search for "3D" should generate a regular SERP. (Costco does this correctly, including both 3D televisions and other products relevant to the query.)
Until people begin to grasp the complexities of search and develop skills accordingly, businesses that take such extra steps to help users find what they need will improve customer success — and the bottom line.