Search Design Best Practices

Search is one of the most important capabilities of every Knowledge Management project on which we work.  Every company of any size has to find a way to make search work.  It can be a website search for customers, a product search, or an intranet that allows employees to find what they need.  Very few of these search tools work as expected.  Most organizations try to solve their search problems with a complicated and often expensive search tool with pages of features.  A full featured search engine provides all of the features necessary to create a successful search, but it alone is not enough.  User experience is the most critical aspect of a successful search solution.   

I have worked on over 50 search projects in my career.  Over time I have identified 5 key design approaches that are critical to a successful search experience:

  • Avoid dead ends;
  • Design actionable search results;
  • Start with a faceted taxonomy;
  • Gather as much information about the searcher as possible; and
  • Plan for change.

 

Avoid Dead Ends

Search screens, like all other areas of your website, should be designed to minimize the number of dead ends that searchers encounter.  Two of the most common causes of dead ends are search results that lack information about the page or document the searcher is going to and poorly designed advanced search screens.

Search results should provide enough information so that the user can make an informed choice about whether they should click on the link.  There is nothing more frustrating than getting a list of search results and opening one link after another only to find out you are at the wrong result. Provide text snippets, images, the author, or other additional context about the search result so that searchers have enough information to decide if they really want to click on the link.  Some more advanced search applications allow for previews of the search result so that the user does not have to click the link to find out where they are going.

Many sites have an advanced search link that opens up a separate screen where the searcher can fill one or more fields to execute a search.  These screens can be a powerful tool but they must be used appropriately.  Not every search requires an advanced search.  The problem with advanced searches are that they can lead to a zero results page.  Because the search criteria is entered up front, the user has no idea why they have no results in their search.  Features like faceting and boolean search allow searchers to filter results one variable at a time to figure out what they are really looking for, without allowing a zero result search.  Another advantage of the faceted approach is that users can discover information along the way that they might not have found otherwise.

 

Design Actionable Search Results

Search is a means to an end.  People don’t search solely to find things. They search for information to support an action they are trying to take.  A well designed search should help searchers quickly take action.

The best way to make this happen is to look at each type of search result and determine what people are likely to do with that information.  Google offers a number of great examples of this approach.  

Restaurants – When searching for restaurants, Google provides a map showing the location of each restaurant, a listing of the restaurants in the area that includes the name, rating, and phone number for the restaurant.  The searcher can easily pick a restaurant and make a reservation.

Stock – When searching for a stock, Google provides a stock chart as well as the current stock price.  The searcher has an update on the performance of the stock in case they want to buy shares.

Retail Items – When searching for things like a TV or a piece of furniture, Google returns results that show a picture of the item, the price, and a rating of the product.  Searchers know what they are getting and can order the product directly from the result.

These examples make great sense for Google, but this same logic can also be applied to business searches.

Conferences or Events – A conference or event search result should include the date, location and a brief description of the event.  Many of our clients also include a link to register or add the event to their calendar.

HR Forms – HR Forms should have a description of the purpose of the form and a link to automatically sign and email the form to the correct person.

Company products – Products should have an image  of the product, pricing information and a link to initiate the purchase process.

People – Search results showing people should include their title, department, email and phone number.  Some companies also include a link to their supervisor.

Actionable search results are easy to implement and they have a huge impact on tolerance towards searching.  Simple design changes to your search results make your searchers feel much more empowered.

 

Start with a Faceted Taxonomy

Faceted navigation is one of the most important features in search.  It allows people to narrow their results and quickly get to the information they need.  Given the importance of facets to the search experience, identifying the correct facets must be a priority.  The facets should be understandable for the user community and they should allow searchers to filter content in a meaningful way.

Work with a Taxonomist who understands how to design deconstructed taxonomies for faceting (as opposed to deep hierarchical taxonomies).  The taxonomy should be done using focus groups to make sure that the terms make sense to your community.  A great primer on how to design a proper taxonomy can be found on our site at Taxonomy Design Best Practices.

 

Gather as much information about the user as possible

Finding what you want in a large corpus of information using nothing more than a single term or a short phrase is unlikely.  It is important to design your search so that it can take advantage of as much information about the searcher as possible.

There are a number of ways to collect additional information about the user including:

  • Location – The searchers location on the system at the time of the search can provide insight into what they are looking for.  For example, an intranet might prioritize accounting material over other material when the search is started from the accounting page.
  • Personalization – If searchers need to log into the site, you can use information about the individual to prioritize certain search results over others.  This is a very common approach in e-commerce sites and company intranets.
  • Searchlets – Specialized searches can be included in different locations on the site so that users are searching a subset of information.  The company contact list is a common example of this.
  • Faceting – As searchers select facets, they provide more information about what they are trying to find.  The search should adjust based on the facets chosen.  New facets can appear (conditional facets) that are relevant to what the searcher is looking for.  Again, this is common in retail.  When someone picks the category of televisions, for instance, additional facets including screen width and LED/LCD technology spawn.  In a standard intranet, a facet of topic/subject is a common means of spawning secondary facets.
  • Query Parsing – Search phrases often provide some insight as to what the searcher is looking for.  Most searches send the search phrase directly to a search engine.  The search engine then tries to find matches.  A good search will look for clues as to what the searcher needs in their search terms.  Google does this well.  Search for “Italian Restaurants” and Google understands that you are doing a restaurant search for Italian Restaurants in the area.  A company search could look for clue words like “Policy” and automatically prioritize content tagged as policy content.

The more you know about your searcher; the easier it is to deliver relevant results to them.  The approaches above can be used to collect a lot more information that will allow you to better prioritize search results for your searchers.

 

Plan for Change

Search should evolve as you learn more about your searchers and your content grows.  The key for long term success with search is to create a solution that can change over time.  Develop a governance process that is driven by analytics.  Also make sure that the development team architects a search that can be enhanced through an administrative console rather than custom development.  

We suggest that our governance teams meet no less than quarterly to discuss enhancements to search.  The team should have a set of reports that help them understand how to improve search tools.  We frequently recommend the following reports:

  • Most common search terms;
  • Searches with no results;
  • Average search result selected;
  • Most common facets selected; and
  • Highest average search result selected by search term.

These reports can be used to determine what people are searching for and which facets are useful.  The governance team should be able to recommend changes to the system based on these reports.

Administrators must be able to enhance search without the need for development or reindexing.  One simple way to affect change in the search system is to allow administrators to “promote” search results.  The Search governance team can look at the most common search terms and promote the appropriate content for those results.  This approach gives the business the power to affect change in search.  It is a powerful way to show that search is in a state of continuous improvement.  However, this too should be managed by clear governance processes. Overuse of “promoted” terms becomes an administrative burden and can potentially corrupt the sanctity of search results.

 

Let us Help

Search design is critical to delivering a solution that your users will find valuable.  These 5 search design best practices will change the way you think about search and how your searchers feel.  If you are unhappy with your existing search solution, we can help get things on the right track.  EK offers a search workshop that will produce a set of recommendations to get your search performing again.  Please contact us for more information.

Joe Hilger Joe Hilger Joe is Enterprise Knowledge's COO. He has over 20 years experience leading and implementing cutting edge, enterprise-scale IT projects. He has worked with an array of commercial and public sector clients in a wide range of industries including financial services, healthcare, publishing, hotel and lodging, telecommunications, professional services, the federal government, non-profit, and higher education. Joe uses Agile development techniques to help his customers bridge the gap between business needs and technical implementation. He has a long track record of leading high-performance professional teams to deliver enterprise-level solutions that provide real value. His development teams have a strong record of client satisfaction, innovation and leadership. Joe is an expert in implementing enterprise-scale content, search, and data analytics solutions. He consults on these areas with organizations across the country and has spoken on a wide range of topics including enterprise search, enterprise content management, big data analytics, Agile development and content governance. More from Joe Hilger »