The Five Cs of KM: Curate, Part 1—Search Results

This begins the second series on knowledge management in special libraries and information centers, structured around Five Cs: Capture, Curate, Connect, Collaborate, and Create. This four-part series is about the second C: Curate.

Curation is taking existing information and making it more useful. This includes better organizing it, making it more findable, and making it easier to use. The techniques for doing so include creating best bets, publishing frequently asked questions (FAQs), and monitoring and enhancing online discussions. Subject matter experts within special libraries are uniquely qualified to undertake this kind of curation.

Three key types of curated content are search results, FAQs, and online discussions. In this first of a four-part series, I discuss curating search results.

Knowledge managers are frequently asked, “Why can’t our enterprise search be more like Google?” While it’s impossible to match the scale of the Internet inside a single enterprise, and thus Google’s PageRank algorithm won’t work the same way, it is possible to emulate other Google functionality.

Beyond organic search results, useful content can be provided more intentionally. To do so, determine the topics of greatest importance to the organization, curate a list of relevant content that can be searched and filtered, and feed the entries as enterprise search results. These can be in the form of curated search results: best bets (thumbnails and links only), authoritatively badged content, or quick answers (more complete content plus links) for the content deemed to be the best for each of these key topics. They can also be dynamically generated using attributes, tags, sorts, filters, human interaction, etc.

Knowledge managers should regularly conduct internal and external searches on topics of interest to users and in response to queries. Select the most relevant and valuable results, compile them, provide them as answers, and publish them in posts and as web content. Users can participate in search result curation by tagging content with “I reused this.” And their actions can indirectly support curation through the use of most-visited webpages and most-downloaded documents in search results.

A best bet is a result that you expect to see when using a search engine or an enterprise search tool. Entering a search can result in a long list of links to documents and webpages. Someone doing a search would prefer to see a recommendation for the content most likely to be useful, so they don’t have to click on all the links or open all the documents. Offering best bets starts by defining the topics that are most important to an organization. This can be done by asking leaders what they think is essential, polling users on what they most want to find when searching, and reviewing enterprise search logs for the most searched-for terms and the most clicked-on search results. Using this information, curate a list of content that applies to those topics. Then when someone does a search, instead of relying only on the search engine and its crawl of available content, feed the curated entries as part of the results.

For example, if someone searches for “vacation policy,” enterprise search may return many documents containing the word “vacation,” some of which will not be relevant. Given that vacation policy is something that people tend to search for, the official policy document can be predefined as a best bet, flagged with a special icon, and offered as the first search result. This allows users to get the information they are most likely to need right away, without any wasted time or effort.

For the best content for each of the key topics, provide thumbnails that provide a glimpse of the information up front before having to click links. Offer authoritative badges using a special star, check mark, or other icon to denote content recommended as being the best or the most reliable. If a user is looking through ten different pieces of content that show up as search results, and one of them has the authoritative badge, they can view that one first.

The most extensive form of search results are quick answers. These have become ubiquitous in public search tools like Google and Bing, reducing the need to click through any of the search results. Essential information is provided right on the search results screen as quick answers, also called knowledge panels. For example, if someone needs to find information on one of the leaders of the organization, a search will result in two or three paragraphs being displayed at the top of the search results. There is no need to click any further.

Here are ten categories for the curated list of quick answers:

1. Internal organizational structure, e.g., Finance or Human Resources
2. Formal taxonomy: industry or internal, e.g., Global Industry Classification Standard or enterprise taxonomy
3. Products and Services, e.g., Android or consulting
4. Topics, e.g., security or supply chain management
5. Industries, e.g., electronics or pharmaceuticals
6. Clients, e.g., GE or US Government
7. Partners, e.g., Ford or Microsoft
8. Locations, e.g., Latin America or Detroit
9. Specialties and Roles, e.g., project management or information architect
10. Demographics, e.g., new hires or retirees

Here is how to choose search results to provide:

1. Request submissions from content providers but be selective in what you accept. Some providers will submit endless lists of best bets, and not all these will be worthwhile.
2. Match the existing intranet high-level navigation so that for each link in the top navigation and/or breadcrumbs, there is a corresponding quick answer.
3. Ask key contacts for each category what they consider important. These can be organizational leaders, thought leaders, subject matter specialists, community leaders, or knowledge managers.
4. Ask in Enterprise Social Network (ESN) groups and communities to get the suggestions of members. Share the proposed quick answers there to get confirmation or changes.
5. Review enterprise search logs for the most searched-for terms and the most clicked-on search results.
6. Review web analytics for the most visited pages and downloaded documents.
7. Review websites and documents that are liked or tagged as useful by users.
8. Ask all help desks to provide the most frequent queries and replies.
9. Review queries posted in the ESN, email messages sent to distribution lists, and requests sent to official mailboxes. Look for patterns of missing or hard-to-find content.
10. Review all published FAQs.