For a recent client presentation, we discussed at length issues the client had with their search engine and poor performance reported by their users. Essentially, users were not able to find the information they required quickly – within the first 3 page results.
The problem specifics are:
  • The search engine would respond with several pages – usually to many
  • Users are not able to spot the item they want – encryptic titles and subjects appear in the results
Sound familiar? To help me understand the sort of work they’ve performed to design and enforce the search functions, I asked a few questions such as:
  • Do you have a common vocabulary? Information Architecture? Sort of the same thing?
  • Do you enforce the common vocabulary through training, mandatory fields etc…?
  • Do you monitor search queries and work to improve Meta data tagging?
  • Do you use synonyms to address variations in vocabulary?

In most cases the answers were no. With clients such as this I generally walk them through a process whereby they learn why their search engine isn’t performing – from a user perspective.

Typically I define what a common vocabulary is – the organizations lingo. For example, every profession has a common vocabulary such as doctors, consultants which are profession based lingo and there are company based lingo – speech community or Jargon as WikiPedia defines it.

For example, what does Oncology refer to? Pediatrics? If you’re a physician you know exactly what that means.

Historically, common vocabularies have been addressed by using taxonomies and various tagging exercises. Experts in this space have argued that taxonomies are difficult to design and maintain. I equate taxonomies to boiling the ocean – something that is almost impossible to do. Case and point, a customer of mine was designing and deploying a portal. The portal was built and deployed…eight months later they still didn’t have a taxonomy. What they did have is what appeared to be communities of practice and the typical company marketing and communications. This leads to my next point on defining communities of practice.

Defining communities is a much easier approach to developing an information architecture. Typically communities centered around areas of practice, organizational functions and interests such as sports and hobbies. For example, a company might have communities such as SharePoint Technologies, Marketing and Mountain Biking.

Once communities are defined, you then determine what information is required such as links, artifacts (documents, images etc…), discussion topics etc… Once you have the artifacts defined you then use the Dublin Core Meta Data standard plus your own companies Jargon to label and tag the artifacts.

To round out the approach you develop test cases to determine how efficient your tagging system is by rating how quickly users can find data. Finnally, ongoing refinement using test cases, usage reports, user training and awareness and client feedback really helps to refine the search engines performance.