People often mispel words when using search engines. Google has registered 593 ways of spelling Britney Spears.

Links:

• 593 ways of spelling Britney Spears

Henrik Olsen - January 26, 2006 - via justaddwater.dk

See also: Misc humor (3) 

According to Jesper Ronn-Jensen from justaddwater.dk, live search will gradually replace traditional search on the web. In live search interfaces results are fetched whenever the user stops typing for a brief moment. An example of this is Google Suggest where the most popular results are presented as-you-type.

Jesper sees the following benefits:
- The search user interface is identical to traditional search
- Misspellings can be corrected immediately
- Relevant alternatives are presented as you type
- It's easy to refine your search: Just continue typing
- If the search is too narrow it's easy to press backspace and remove characters

Links:

• Live search explained

Henrik Olsen - January 26, 2006

Much can be done to improve the quality of search results. But according to James Robertson, no amount of tweaking search engines will ensure that the most relevant results always appear at the beginning of the list. This is where "best bets" come in.

Best bets are a hand-created list of key resources for common queries, presented prominently at the beginning of the search results. By analyzing search statistics, we can ensure that the most useful pages are listed right at the top of popular searches.

Links:

• Search engine 'best bets'

Henrik Olsen - December 06, 2005

See also: Tips and guidelines (63) 

Based on a usability test of a system that allows people to search a large set of content Donna Maurer interpreted the users' expectations of search:

- It is better to put more than one word in as one word gives too much stuff
- Adding an extra word gives fewer results
- The first word in the search box is more important than the other words
- If the words make a sensible phrase the search engine should return results for the phrase
- If the words do not make a sensible phrase, the search engine shouldn't look for the phrase.

Links:

• Regular folks searching

Henrik Olsen - October 14, 2005

See also: Research (91) 

It's time for Jakob Nielsen's Top Ten Web Design Mistakes. In 2005 Jakob has asked his readers about their opinion. Here's the result:

#1 Legibility problems due to small fonts and low contrast
#2 Non-standard links that violate common expectations
#3 Flash with no purpose beyond annoying people
#4 Content that is not written for the web
#5 Bad search
#6 Browser incompatibility
#7 Cumbersome forms
#8 No contact information or other company information
#9 Layouts with fixed width
#10 Photo enlargements that doesn't show the users the details they expect

Links:

• The article Top Ten Web Design Mistakes of 2005

Henrik Olsen - October 03, 2005

See also: Forms (11)  Text (14)  Links (10)  Flash (6)  Browsers (2) 

37signals have made their e-commerce search report from 2003 available for free. The report looks at the usability of search results from 25 of the internet's leading online retailers, and concludes with a comprehensive set of best practices.

For each retailer 37signal have tested:
- Are the search results accurate and relevant?
- How does the site handle misspellings?
- Can I sort the search results by useful criteria?
- Will the site understand related words and common synonyms?
- Can I search using mixed specifications such as gender, color, and price?
- Does the site provide helpful tips when it returns no results?

Links:

• The report Evaluating 25 E-Commerce Search Engines

Henrik Olsen - August 15, 2005

See also: Tips and guidelines (63)  E-commerce (21) 

According to Jakob Nielsen search is such a prominent part of the web experience that users have developed a precise idea of how it's supposed to work. Deviating from users' expectations almost always causes usability problems.

Users expect search to have three components:

- A box where they can type words
- A button labeled "search" that they click to run the search
- A list of top results that's linear, prioritized, and appears on a new page

Given how ingrained it is, it's crucial to avoid invoking user's expectations for other interactions. Users' expectations are so strong that the label "Search" equals keyword searching, not other types of search.

Links:

• The article Mental Models For Search Are Getting Firmer

Henrik Olsen - May 09, 2005

See also: Tips and guidelines (63) 

This paper describes an interesting study of e-commerce sites that was set up to determine factors involved in the decision to use search or browse menus to find products.

According to the authors Michael A. Katz and Michael D. Byrne, the decision of a user to search or browse a site is affected by multiple factors including:
- The site information architecture in terms of labeling and menu structure
- The user's inclination to search
- The prominence of search and browse areas

They found that:
- Given broad, high-scent menus, participants searched less than 10% of the time, but they searched almost 40% of the time when faced with narrow, low-scent menus
- Participants showed a higher success rate when using the menus to find products as opposed to search
- Searching for products wasn't faster or more accurate than browsing

Links:

• The article Effects of Scent and Breadth on Use of Site-Specific Search on E-Commerce Web Sites (PDF)

Henrik Olsen - February 24, 2005

See also: Navigation (44)  E-commerce (21)  Research (91) 

According to Daniel E. Rose, current search interfaces reflect the inner workings of search technology rather than what we know about how people look for information. In his opinion, we should use our understanding of search behaviour to rethink how we interact with search engines.

Search interfaces should be guided by three principles:
- Provide different forms of interaction to match different search goals
- Facilitate selection of context for the search
- Support the iterative nature of the search task

Most of the time, search is an iterative process like the interaction between a reference librarian and a library patron. Users don't know the right questions to ask until they begin to see some of the results and learn about the subject. Rose mentions the AltaVista Prisma feature, which suggests search refinements, as an example of how search engines can support this iterative process.

Links:

• The paper Reconciling Information-Seeking Behavior with Search User Interfaces for the Web
• Background info on AltaVista Prisma

Henrik Olsen - November 08, 2004

See also: Tips and guidelines (63) 

The authors of this article have studied the optimal layout of search result pages. Their findings suggest that categorizing search results improve users' performance significantly.

The authors tested seven different search result layouts, and found that:
- In all cases, categorized search results were faster than non-categorized results
- Despite the cost of additional scrolling, the layout with categorized search results, page titles and text summaries was the most effective
- Participants generally preferred the categorized results to the non-categorized
- Adding category information to non-categorized results didn't improve performance
- Removing category names from categorized results didn't hurt performance, but the participants disliked the absence of a category name

Apparently, categorized search results help users weed out irrelevant results and focus in on the area of interest more quickly.

Links:

• Optimizing Search by Showing Results In Context (PDF)

Henrik Olsen - October 10, 2004 - via Semantic Studio

See also: Research (91)