There are many ways to get to content on the intranet: navigation, cross links, tags, via employee profiles and... search.

Most of us use Google to search on the internet and we are quite happy with the returned results. But often we are not so happy with the search in the intranet.

Intranet search is still a feature that leaves a lot to be desired in many organisations.

This article contains some lessons I learned implementing search in an intranet with the Google Search Appliance¹ and how to prepare the intranet content for it. There are also many tips you can use without the GSA. We took search serious and it turned into a mini project next to delivering the intranet.

Do we need search?

Gerry McGovern² says "I only use search when I can't navigate". I agree that the site navigation is the primary way to get to desired content. However search can be very useful to search in archives, in documents or other content that is not so obvious to place in the navigation. And some people just simply prefer search!

With the rise of social elements in the intranet, the intranet can become a mash up of different integrated systems. There is still the traditional content management system, but also applications for the wiki, the blogs or microblogging. A good intranet has one search feature that returns relevant results of all integrated platforms.

Improving Search

There are four main areas in which the intranet search can be improved:

• Improve the search interaction
  Help the user finding the information it is looking for. Help them to find information that they otherwise might miss e.g. by giving feedback on the search terms.
• Provide a great user interface
  Improve the layout and the items returned by the search
• Write good quality content for the web and improve the quality of the indexed data
  Write content that is well understood by the search engine, but make no sacrifices to the readability by human visitors
• Monitor and evaluate the search
  Get feedback from the visitors e.g. if they found what they were looking for

 

Improve the search interaction

1. Spelling suggestions

Benefit: Gives the user an opportunity to run the search again with a corrected search term.

When a visitor types a search term that seems to be a misspelling, a spelling suggestion will be shown. For example, if a visitor types "prising," the search engine will respond with the following spelling suggestion: Did you mean: pricing (The decision of clicking a spelling suggestion is of course up to the visitor).

2. Synonyms

Benefit: Increases the number of relevant results for a search query.

The query is automatically expanded by adding synonymous terms.

For example, a visitor searches on the term "documentation". The search engine will return the most relevant results that contain the keyword "documentation" But the visitor will miss results that contain alternative terms, such as "guide" and "manual." If the search term was expanded to include "guide", "manual," and "manuals," the search query would be wider and would return more relevant results. Other examples could be 'HR - Human Resources' or 'Salary scales - Salary matrix'.

3. Search suggestions

Benefit: Helps visitors refine their searches by offering related queries.

Suggest an alternative word or phrase for the user's original search term.

For example, a visitor searches using the term "Global warming" which is popular term for "Climate change" Searching for will return results indexed under "Global warming" but misses the results indexed under the new term. With this technique the search engine will suggest the following related query at the top of the search results: You could also try: Climate change (When the user clicks the suggested search term, the query will be rerun).

This technique differs from synonyms that it can be used to make employees aware of changed terms within the company!

4. Direct matches

Benefit: Guides visitors to recommended links.

Direct matches are recommended links that appear at the top of the search results and give visitors an opportunity to navigate immediately to the recommended page or document. These recommendations can be specified for specific search terms. 
Direct matches can be used to promote specific content or documents, even when this content is not indexed or has a low calculated relevance based on the search term.

5. Real time search results

Benefit: Provides real-time access to data from external sources

In some instances, the most relevant result for a search query is real-time data. This type of data does not usually reside in the search index because it would be obsolete before it could be indexed.

Some examples you can think of are results from activity streams (microblogging) in your company or calendar information from Microsoft Exchange. But also results from external sources like flight information, stock prices.

6. Filter on Metatags

Benefit: return only search results in a specific area.

Any intranet should have well chosen content types and metadata to classify the content. This can be used to link content to each other on the front end, but this is also very useful to use when searching for content. Provide an option to filter on the metadata to narrow the search per topic: e.g. search in procedures, manuals, press releases, IT content or committee minutes only. Filters also can be applied to: Language or file type.

Convert all metadata in the CMS to HTML Meta tags on the presentation side (the html page in the intranet). And provide an option to let the search engine filter on these tags.

7. Mix the search results

Tip: You could think about combining the search with results from the organisation customer facing websites.

Provide a great user interface

There are various aspects of the search user interface that can be improved to meet the specific needs of your business or visitors.

1. Layout

Best practices tips and guidelines for designing an effective user interface:

• Keep search pages clean, simple and fast 
• Keep advanced search separate 
• Make search available on all pages
• Make sure the search box is big enough
• Make help easily available
• Keep the number of results reasonable

 

2. Hide duplicate results

Benefit: Limits the number of returned results by only showing the most relevant results.

Present only the most relevant search results by eliminating duplicates. Google uses these filters to eliminate duplicate search results

• Duplicate Snippet Filter
  If multiple documents contain identical titles as well as the same information in their snippets in response to a query, only the most relevant document of that set is displayed in the results.
• Duplicate Directory Filter
  If there are many results in a single web directory, then only the two most relevant results for that directory are displayed.

 

3. Provide Useful search help

Benefit: Teach the users how to get more out of search.

Write help on how to search the search. Each visitor is unique, and the search engine is not (yet) smart enough to fully understand what the visitor exactly means. Therefore some explanation of how the search engine works is helpful.

4. Highlight the searched words in a PDF document

Benefit: A user does not have to search the PDF document again (less clicks, better experience).

When clicking a PDF from the search results open the PDF document in search mode to highlight the searched words or jump to the desired page in the PDF.
All what is required is a small modification to the URL to the document.

5. Make use of the Robots.txt file.

Benefit: Do not show totally irrelevant results, don't even index them.

To prohibit any crawler from accessing all or some of the content on the intranet you can add a robots.txt³ file to the content server or Web site. This file tells the crawlers to ignore all or some files and directories on the server or site.

6. Don't index certain page elements

Benefit: Do not show false search results.

Some items on a page should not be indexed or should at least not be returned in the search results:

• The publisher of the page (NB the publisher is here not the same as the owner of the content)
• Navigation (Don't return pages that are triggered by text in the global navigation)
• Lists (dynamic and static) that refer to other articles. (This is the reason why Google gives pages with more links outgoing a lower Page rank than pages with more links coming in).

 

Write good quality content for the web and improve the quality of the indexed data

1. Writing for the web and SEO

 

When an article is about 'Bulk Information', but never mentions the words 'Bulk' or 'Information' in the text, it will be hard for the search engine to return it when searched for 'Bulk Information'.

SEO rules count for the intranet as well, but never make sacrifices to the readability for human visitors

• The meta tags keywords and description have little impact on the Enterprise Page rank calculated by Google. Content is King now. (This follows the theory that the best content will generate the most links to it and therefore have the highest Page rank and therefore will be on top of the search result page). But on the intranet you are probably not bothered by keyword spammers, so you can use this powerful technique.
• Keyword density has a high impact - use synonyms in your content. Think about how people might search for the content
• Try to use HTML tags as they should be. An address is a <address>, a table is a <table>, a heading is a <h>, a abbreviation is a <abbr>, a paragraph is a <p>, a list is a <li> and so on.
• Use descriptive urls - an article about shared principles should be called 'shared principles'
• Do not use a single directory for all documents, but use separate folders for your content. Store them by topic.
• This is just a shortlist; there are many articles on the internet about 'writing for the web'.

 

2. Get the dates of documents right.

Benefit: Sorting search results on date shows the correct result.

To enable search results to be sorted and presented based on dates make sure you use the correct date in a format that is understood by the search engine and provide them where applicable: e.g. in the title, in the body of the document, in the metadata of the document.

3. Use search often and creative

Use a search boxes on specific topic pages, on unit pages or any page that shares some common metadata. Use these search boxes (with prefilled filters) to find content about similar topics.

4. Metadata (filter on meta tags)

Any intranet should have well chosen content types and metadata to link content to each other, but it can also be used to search for content.

The usage of right metadata in the intranet is very important. It can be used to link content to content and content to people.

5. Feed the search engine with metadata on binaries.

Benefit: To filter documents the same way as with HTML web pages

Documents are not in HTML format and can therefore not be filtered with the use of html meta tags. With the GSA it is possible to push metadata with binaries. This makes it possible to return combined results from documents and web pages!

Monitor and evaluate the search

1. Get Feedback

Benefit: Learn about the behaviour of users and be able to improve the search based on their feedback

User feedback is also an effective way to identify usability issues.

When you solicit feedback, it is important to let the employees know that their opinions about their search experience are important. As they see improvements based on their feedback, they will trust the search feature more.
Some easy examples to enable users to give feedback:

• Add an email link to the search results page. You could e.g. label the link "Email us if you can’t find what you’re seeking"
• Add a button on the search results page. You could label the button "I didn't find what I was looking for" Log the clicks with the corresponding search query.

Feedback that is sent with information of the user is very useful. It enables the possibility to contact users and to get more information and clarification and show that you are following up with them.

A survey could be an effective way of soliciting feedback from users about their search experience as well.

2. Analyse the search requests log file.

Obviously it is very important to log the search queries.

A good way to monitor the search behaviour is examining the search log files for the most used keywords and search phrases. It is especially important to look at search phrases and keywords that did not return results. Analyzing the log files helps you find the most popular items on the intranet. No... this is a JOKE! On the contrary: It helps you find information that is not easily found by other means or information that is not even present on your intranet!

Analyse the search logs and ask yourself why people are using search to find this information.

• Analyse search requests that returned results.
  Use this to check if the right results is in the top result page
• Analyse search requests that returned no results.
  Ouch, this means that people are looking for something that is not indexed right or worse... that is not available on the intranet.

 

Some notes

 

Crawl, link texts and page rank, index documents

The GSA builds its index by crawling the intranet pages. It indexes web pages and also the documents (word, pdf, xls, ppt) stored on the intranet. That means it follows links on the intranet. It also uses the text in the link to determine what the linked page is about. It counts the number of links that point to an article and uses it to calculate the Page Rank⁴ The more other pages link to a piece of content will increase the Page rank of that article. Articles with higher Page Rank will be placed higher on the result pages. It calculates the relative importance of pages in combination with the search terms).

However, we should be aware that the Page rank calculation depends on a high number of cross links. This is not always the case in the intranet. How more cross links there are, the better it works.

Internet Search vs. Intranet Search

There is also a difference between search on the internet and the intranet. On the intranet you expect certain search results, because you know that the content is published on the intranet. You don't know the same thing on the internet. In other words you won't miss it in the search engine result page. People are more critical with intranet search results.

 

You can follow me on Twitter at @BasZurburg.

This article as PDF: Creating the Intranet Search Experience.pdf (131.58 kb)