Improving Website Performance via Advanced Tagging

Practical guidance to dramatically improve browsability

Part X of this series of articles that share useful insights and practical guidance to troubleshoot underperforming self-help and learning systems

As our intranets, extranets, marketing and self-help websites scale, the nature of both content and demand will change. With the need for flexibility, problems can arise when organizations opt to invoke a rigid classification scheme as the basis for perusing online resources. This article examines how simple tags (labels) can be adapted for intuitive and flexible browsing.

There are many significant benefits arising from the consistency of a controlled vocabulary. With intuitive, well-conceived classification, one can set up a wonderfully organized set of virtual bookshelves for locating choice resources. Perhaps most importantly, classification is the foundation for cross-referencing. For example, a query for Travel + Form will surface all travel forms.

With a classification scheme in place, it is a trivial matter to label any new resource for easy discovery…until it isn't anymore.

And there’s the rub; inevitably, the terminology used by our users evolves until the antiquated vocabulary of the classification scheme is more a hindrance than a help. The obsolescence is built in.

Consider:

• Emerging needs: Nostradamus wouldn't be able to predict the changes in demand for online resources over time. No scheme devised in years past will remain aligned with the real world absent recognized authority and user buy-in. And even then, it’s likely to be a painful experience, e.g. pharmacology’s reliance upon Latin: hora somni = at bedtime.
• Nuance: Controlled vocabularies are notoriously challenging to configure because of the need to balance experts’ calls for precision with novices’ need for discoverability. For example, the differences between standard, guideline, policy and regulation are stark to some seeking authoritative information, but maddeningly similar to others.
• Evolving meaning: As new terms emerge, there is a period of uncertainty during which usage and meaning can change.
  Consider webinar and long form.
• Competing/conflicting sources: Like it or not, the methods other online support systems use to classify their resources can set the expectations of your users. If amazon.com uses the category Patio & Garden, you can bet that Target and Sears are stuck with it.

About a decade ago, some bright folks tackled these problems and developed the concept of tags. Tags are simple labels that are nonhierarchical and very flexible. They can be applied by users and publishers in a manner that democratizes classification.

With ubiquitous tagging, anyone can act like a librarian; one can apply tags via your operating system, email, online discussions, online retailers, even your bank probably offers tagging. But do you tag?

Unless your spice cabinet and pantry are categorized, you probably do not go to the trouble of adding tags, even though the potential benefits are significant. It’s a practicality issue for most; it takes time and effort to consider the myriad options, and it is difficult to remain consistent over time. Moreover, it’s often faster to conduct a quick search than it is to browse via informal tags, even within your own knowledgebase.

Tagging and controlled vocabularies are both valid concepts, but each has a serious weakness:

• Tagging doesn't scale well
• Controlled vocabularies become obsolete

Can we combine the flexibility of tags with the
utility of controlled vocabularies?

Experience suggests we can. Consider these insights when devising a classification scheme:

• The purpose of tagging is navigation, not organization. Tags are not intended for content management (as are folders); they are labels to assist the user in discovery.
  Note: Expect colleagues who contribute to the knowledgebase to naturally confuse the two.
• Hierarchy is necessary. One must be able to group/nest so that the user can readily peruse the options via drill-down. If everything is at root level, it becomes un-navigable as the system scales. So, even though the original concept of tags was nonhierarchical, ignore that well-intentioned but overly restrictive constraint and group the tags into categories that are intuitive to your target audiences, e.g. a colleague at your company might naturally look for information about benefits under the category of HR.
  Note: Strive to limit the number of top-level categories to 8-12 total to maintain navigability. It is wise to build in flexibility so that tags can be parented and re-parented or exist at root level as needs change.
• User Orientation. A common mistake is to label the resource according to functional ownership rather than use. For example, the legal department may be responsible for maintaining a much-needed form, but the users may be looking for it elsewhere by topic, e.g. travel.
• Editable labels. A controlled vocabulary need not mean a fixed vocabulary. As the terminology morphs, adjust your labels accordingly.
  Note: One solution is to have both a display value and a identification value for each tag. One can then edit the display value to be anything it needs to be.
• Blend tag types. It is extremely useful to have a set of complementary tag types, for example, topic, type of resource, and region. This is a powerful and intuitive way for users to quickly locate and display relevant material. It’s also an effective way to ensure authoritative results. Combining tag types such as Construction Safety + Learning Resources + Asia/Pacific is much more effective and definitive than expecting the user to cull through resources with only topic as a differentiator. With a blend of tag types, the user can quickly and confidently determine whether there are any training materials related to safely using bamboo scaffolding.
• Content before tags. Never create a tag in anticipation of content; always require that resources be available to spur the creation of tags. This simple rule helps keep the tag scheme uncluttered and navigable. Moreover, absent content to guide labeling, important decisions will be based on assumptions about the nature of the resources. In practice, do not differentiate a group of resources with tags until the number of resources approaches unwieldiness. Put another way, tags with one or two resources are a symptom of over-classification.
• Don’t over-tag. A common mistake is to label a resource with all tags that apply rather than using only the tag or tags that best apply. Over-tagging frustrates browsers by forcing them to encounter the same items no matter how hard they try to differentiate them. Another common error is to apply all child tags to a topical resource; for example, an employee manual that covers all aspects of human resources should not be overtagged with child tags such as benefits, vacation, termination, etc. Such a resource should be tagged with the category tag (perhaps HR), and then featured as a prominent resource and described as being comprehensive.
• Define tags. Keep the tags brief, but offer the user an explanation.
  Note: One technique is to associate a description field with each tag and display it via a mouse-over text bubble. This field will describe what the user should expect to find.
• Display in context. One can accommodate tag browsing by using best practices such as indented hierarchy and making the selections bold to help users grasp where they are in the knowledgebase.
• Quality control. Permission to apply tags should not be given lightly nor without training and oversight. Topical subject matter experts and departmental authorities (HR, legal, IT) should be engaged for reviewing proposed tags in their domains. If the organization has a support center or help desk, your colleagues there are a terrific source for feedback as new labels are tested and refined.
  Note: Poorly configured metadata is a quality control problem that is surprisingly difficult to expunge.

Classifying content to enhance browsability is both challenging and rewarding. You are the user’s advocate with the responsibility to make judgement calls that affect people seeking important answers. While working on a dynamic knowledgebase, you can make more meaningful decisions in a day than many of your colleagues make in a month. You’ll need flexibility as your system scales and content changes; tag browsing is an option worth considering.

Next Topic: Adding Externally-Produced Content

Part XI offers insights from the author’s experience helping organizations overcome costly (and avoidable) mistakes when acquiring and integrating commercially produced knowledgebase content, e.g training, professional development, self-help.

Directory of All Topics

Browse this directory to discover how to troubleshoot the often thorny problem(s) preventing our self-help, intranet, training, support and/or extranet knowledge bases from being incredibly useful.