Limiting Search Boundaries

Scoped search allows users to limit their search to a section or type of content on a website instead of searching everything in one go.  Typically, it is implemented in two ways: drop-down scope selection and autocomplete scope suggestions, which appear below or within the search box. 

Drop-down Scope Selector
Brookings.edu: Drop-down scope selectors included within the search box require users to open the menu to see the options available.

 

Autocomplete Scope Suggestions
Google Shopping Express: Autocomplete scope suggestions: as the user types the search terms, categories are automatically suggested just below the search box.


E-commerce sites often use both implementations in the same search box, while information-based sites tend to use one or the other.

A third, less common approach to limiting search boundaries uses two search boxes on a page, each with a different scope. (This implementation is consistently problematic; the only instance where our research has shown it to be effective is for employee directories on intranets.)

Two search boxes
Hunter.cuny.edu: Having two search boxes on a page creates significantly more work for visitors, who, assuming that they notice the existence of two search boxes, must stop to determine which box to type in. Many users will just pick one and hope for the best.


This article presents the main issues with scoped search, as well as recommendations for getting around some of these issues.

Good Intentions with Disastrous Consequences

Websites implement scoped search because it has the potential to help users find what they are looking for faster. At least in theory, scoped search saves users a step by narrowing their search up-front (before the results page) and listing only items that fit their desired categories. With this proactive approach, irrelevant results are excluded, and users find their answer sooner. Sounds great, doesn’t it? Not so fast.

When a user doesn’t realize that his search is limited to a section, the consequences are devastating. At best, users will scour the list of results and struggle with different keywords to produce the desired result, inevitably becoming frustrated and leaving with a bad impression of the website. At worst, users will assume that the website doesn’t have the answer or the product they’re looking for, and they will abandon the site (and likely go to a competitor).

Below are 5 main issues with scoped search.

1. Users expect search to include the entire site.

People expect to be able to enter a term in a search field and get relevant results from anywhere on your site. To most people, anything on the website is part of a single entity, and search should include all of it.

Users’ understanding of the ‘entire’ site depends on their mental model of the website or organization. For example, some visitors to the website for the University of Pennsylvania’s business school (The Wharton School) might think of the school as the University of Pennsylvania. Others might think of The Wharton School as its own entity. Depending on the visitors’ mental models, their expectations of whether search will include the entire university’s website, or just the business school, will be different. For your own sites, figuring out how your users understand your content requires a mix of UX research methods, but examining your site’s search logs is an easy place to start.

Sites that select a scope by default are the worst offenders, because users have not made that choice. Most people will not take the time to study the interface to figure out what is included or excluded, because it doesn’t even occur to them that something might be excluded.

Even customers that do use scoped search frequently forget that they’ve limited the search. So when those users return to the search box, they often overlook the previously selected scope. Particularly on websites that offer a large variety of products, people may do one search after another without remembering or realizing that they are in a scoped area.

Recommendation: Always set the default scope to ‘all’. Any scoped search should be set to the broadest, most lenient category, which should include your entire site. Search should include links to informational content (delivery options, return policy, store hours, contact information, etc.) in addition to product offerings.  

Search scoped by default
Sainsburys.co.uk (one of the largest supermarkets in the UK) defaults its search to Groceries, which is likely to go unnoticed. Even worse, the drop-down scope selector only has two categories, which do not represent the full range of products that this company sells. Shoppers looking for other items or for information such as store hours are simply out of luck. 

2. Users overlook widgets that act as scope selectors (e.g., drop-down menus or autocomplete suggestions). 

People are task focused: they think about the end result, not about what the interface allows them to do (or what the system has selected on their behalf). Most often, they start typing in the search box and click Go without bothering to look around for drop-downs or autosuggestions and to make sure that the default selections are right. Users expect your search feature to behave like tons of other search tools that they’ve seen on the web. And most of these other search tools are not accompanied by widgets that need to be properly initialized.

Recommendation: Make scoped search apparent by using strong visual cues. Placing the scope label in close proximity to the search box and at the top of the search results will help make a scoped selection more obvious. (But even so, there’s no guarantee that users will notice it. That’s simply the nature of designing interfaces: humans have limited attention spans, poor research skills, and a lot already in their working memory.)

Also make sure that the SERP (search results page) states any scope limitation clearly (e.g., only searching ‘music’) and offers a one-click option to expand the search to the entire site.

3. Scoped search forces people to make a decision too soon.

Asking a customer to decide in advance what category they want to search for increases their cognitive load by causing them to make a decision. But people want to do the least amount of work possible. That’s not a criticism: our brains are designed to work efficiently, devoting the minimum resources necessary to accomplish a task, so that we can focus on everything else that’s going on in our environment. (You can learn more about how humans process information in our course on The Human Mind.)

Users behave like stubborn, bullheaded teenagers when it comes to searching for information or products online. We do what we want and deal with the consequences later. That is why most people who use scoped search only get to it after they’ve seen the list of results and realized that it does not meet their needs. Only once they’ve seen that the system hasn’t understood their search query do people begin to think about how to clarify their search terms and narrow their results. Users are more likely to change the search scope in retrospect than during their first search attempt. Let users make decisions about how to refine their search after they’ve seen the initial result list.

First suggestion is not scoped
Target.com: In this example, it’s good that the first item in the autocomplete list doesn’t have a scope applied to it. A downside to presenting scope suggestions for similar categories is that it causes the user to question whether by selecting one of the suggested scopes, she will miss out on any products.


Recommendation: The same recommendation made under issue #1 applies here: Always allow users to submit a nonscoped query. Have the default item in the autocomplete suggestions be without a scope. Thus, you ensure that users only end up in a scope if they have directly taken that action.

4. Scoped search only allows single selection of categories, when multiselection might be more appropriate.

When people want to search more than one of the suggested categories, they don’t have the option. Similar to the previous point, this increases the cognitive burden, as users are forced to wonder how they can select two or more scopes.

Recommendation: Provide obvious ways to refine search results and expand the scope. Evaluate if your users truly need to select a scope up-front, or if multiselect filters (e.g., checkboxes) on the results page are sufficient. (Learn more about designing effective filter and facet controls in our course on Information Architecture: Navigation.) On the results page, users should also be able to remove any selected scopes with one click.

5. Indistinct groupings and unclear labels increase cognitive strain.

Any time users must pause to consider the differences between categories, the interface is making them work too hard. When this happens, it’s often a symptom of bad navigation design or of structural issues with the information architecture of your entire site. Organizations with many divisions and types of content must also remember that people don’t know (or care about) how that company is organized and where information might live among the different divisions. They only care about their goal: to find what they are looking for.

Overlapping scope categories
Categories overlap in the scoped search menu on IMDb.com. What is the difference between Titles, Names, and TV Episodes? How is All different from Keywords?


Recommendation: Search scopes also need to have clear, distinct category names.

To Scope or Not to Scope?

Only use scoped search if you can design it effectively. Scoped search has advantages, but when poorly implemented, it can leave visitors thinking that your website doesn’t have what they want. If your website uses scoped search, keep in mind the above recommendations to minimize risk and maximize findability.