A simple, easy-to-use search tool is critical for a successful ecommerce site. If people can’t find your products, they can’t buy them. A bad in-site search can frustrate users enough to make them abandon you for a competitor.
We recently conducted extensive research on a wide variety of ecommerce sites, looking at different factors that contribute to a positive online-shopping experience. In-site search was an area of special focus for that study. We found that, since our two previous in-depth studies in 2000 and 2011, the landscape of ecommerce search has changed dramatically.
We conducted three studies, each examining many components of ecommerce sites, including search. The studies involved a variety of methodologies, ranging from in-person usability testing and remote usability testing to eyetracking research and diary studies, on desktop and mobile sites.
Higher Search Success Rates
Over the 17 years in which we conducted our three studies, we’ve seen steady improvements in search success.
Search success refers to whether users were able to execute the search successfully and saw appropriate results. Search success does not equal task success, nor does it mean that participants decided to purchase products found through the search. Think of search success as indicating that search didn’t impede users from completing the task successfully — a way of measuring whether search is doing its job.
Between our first study in 2000 and our second study in 2011, overall search success rates increased with 10%. Between our second study and our third study in 2017, we saw an increase of 18%. We observed similar increases in the proportion of searches that were successful on the very first try — participants didn’t have to reformulate their queries. A one-sample proportion test showed a statistically significant difference between overall success in 2017 (92%) to overall success in 2000 (64%) and overall success in 2011 (74%), p < 0.05.
These improvements are encouraging, and are likely due to a variety of factors. Some of the influencing factors responsible for these improvements are not directly related to search — advances in devices, displays, hardware, and connection speeds are helpful in improving online experiences overall. However, some positive trends in ecommerce search certainly play a role in the increased search success. All are covered in detail in our full-length report on ecommerce search (4th edition). Here we focus on four of these trends:
- Refined in-site search and ranking algorithms
- (Somewhat) standardized presentation of search boxes
- Reduced popularity of overly complicated advanced and scoped searches
- Widespread use of faceted search
- Popular use of search suggestions
Improved Search and Ranking Algorithms
Almost all the sites that we re-tested after 6 years showed dramatic improvements in the quality of their search engines and ranking algorithms. This was perhaps the most influential contribution to search success.
We used to recommend supporting shorthand for advanced-search operators, to cater to experienced users. With such shorthand support, a query like “Seagate AND hard drive” would return results that matched both “Seagate” and “hard drive.”
We no longer recommend supporting complex-search operators, because few users know or use them these days. And they shouldn’t have to know them — the search engine should be smart enough to return good results without forcing people to use unnatural language. Perhaps the one exception here is using quotation marks to force the search engine to look for the entire phrase, not individual keywords. That search operator is used by Google, and is commonly known by advanced web users.
One example of an improved search algorithm was on The Container Store’s website. In an earlier study, participants were disappointed by the lack of precision in its search results. One user wanted to purchase a set of stainless-steel canisters with clear tops. When she searched for “steel and glass canister,” she was annoyed to see results like toilet brushes and jars. She tried reformulating her results multiple times, with no luck.
The problem with The Container Store’s earlier search functionality was that it returned all items that matched any of the keywords (“steel” OR “glass” OR “canister”), and didn’t prioritize the items that matched all keywords at the top of the page. Since that test, The Container Store has improved its search by ensuring that those products matching all or most of the keywords are shown first.
Another step forward in search and ranking functionality is the growing popularity of sorting products by a weighted review score, not just by the average review score for each product.
When users sort search results by Customer Reviews, they don’t want to see a product with a 5-star reviews that has been reviewed by only one person. People are very wary of fake reviews, and they’re suspicious of average review scores based on just a few reviewers. When sites sort by weighted reviews, a product with an average of 4.9 stars but 342 reviews will be ranked higher than a product with a 5-star average but only 3 reviewers.
Standardized Presentation of Search Boxes
Most sites are doing a good job of conforming with an external general standard of how to present search boxes and results, but there are some notable exceptions.
The search box is almost uniformly located alongside the global navigation on websites these days. That eliminates many previous issues we found in testing: needing to repeat the search box on results pages, interference from navigation menus, missing search boxes on some pages, and so on. The external consistency of this placement across sites helps users learn where to look for search.
An explicit Search label is no longer necessary (though it won’t hurt). Users look for a wide, empty field or a magnifying glass icon. Many sites use just a magnifying glass alone in mobile designs, and that works fine. Others display an entire search field, which is worth the tradeoff of valuable screen space if you know your sales depend on search. On desktop, sites should still use a simple, visible empty box with a Search button or magnifying glass icon. (Unfortunately, many sites still break that pattern in various unnecessary ways that hurt search discoverability — for example, by using a tiny magnifying glass icon on desktop, even though there’s plenty of room for the full search box.)
Decline of Advanced and Scoped Search
Ecommerce used to rely heavily on advanced searches and scoped searches to help users manage complex product offerings. That reliance has decreased substantially, and with good reason — people don’t use advanced search, and they often get confused and stuck in scoped search.
Now, these complex search features are only present on the sites where they really make sense. Advanced searches are still used on sites with special search needs, like Ebay.com and airlines like Delta. Scoped search is still offered by sites with enormous product offerings, like Amazon and Wal-Mart.
In many cases, both advanced and scoped searches have been replaced by faceted search. The key difference is that faceted search allows users to narrow their options after they submit their query, rather than before.
In our first study, 27% of task failures were a result of not being able to locate a suitable item on the site, even though all of our tasks were designed so there was always at least one item available. When users see too many options, with no way to narrow choices, they can easily miss the products that best fit their needs.
Faceted search allows users to narrow their results set using filters based on the attributes of the products they’re considering. Faceted search used to be a nice-to-have feature on ecommerce websites. Now users notice and complain in the rare cases when facets are absent. These days, ecommerce sites without faceted search are the exception, rather than the rule.
Just as faceted search has become expected rather than exceptional, so have search suggestions. Search suggestions (sometimes called autosuggest) are recommended queries that appear in a dropdown as users type in the search box. When those suggestions are accepted, they can save users time and mental effort, and help them avoid typos or other errors.
In our testing, however, we noticed that, while search suggestions are frequently offered on ecommerce sites, those suggestions aren’t often selected by users. In our third study, search suggestions were selected by users in only 23% of the instances where they were offered.
In many cases, users would just continue typing and submitting their own query. But even when search suggestions aren’t used, they still provide a benefit by giving users an idea of what’s available on the site, and what others search for.
Additionally, we noticed that rich search suggestions (search suggestions that include recommended products, thumbnails, and other content in addition to query suggestions) are gaining popularity on some ecommerce sites.
Rich search suggestions were a brief trend more than five years ago, but faded from widespread use. Now they appear to be having a comeback, taking a form similar to large megamenus.
In our tests, rich search suggestions seemed to work best on sites with many diverse product categories or with products that were very visually different from each other.
Better Search Means Higher Standards
There is a double-edged sword of overall improvement in user experience: As users have more frequent positive experiences, they raise their expectations. As the industry improves, we’re all held to a higher standard. When everyone is capable of providing a functional search, just being functional won’t be enough to set you apart from your competitors.
It isn’t enough to just have a search feature that people can use successfully. You need to provide a seamless, smooth experience, which makes people feel in complete control of their options.
Key Challenges for Search
In pursuing the goal of a seamless in-site search, we identified plenty of common obstacles. Going forward, these will be some of the biggest challenges for ecommerce search:
- Low-visibility search boxes (for example, behind a tiny magnifying glass icon on a large desktop screen, or hidden under a hamburger menu on mobile)
- Lack of support for typos, errors, or common keyword synonyms
- Nonstandard presentation of results (pagination, sorting, filtering)
- Poorly executed filters (irrelevant attributes, bad functionality, empty result sets)
For more details about this study and more tips for creating an exceptional ecommerce search, see our full-length report with 80 evidence-based UX design guidelines for search.