Building a user-centered taxonomy

Your website could contain a large amount of content which covers overlapping, semi-connected and completely discrete subjects.

Connecting these subjects using a user-centered taxonomy will help users find content they want and enhance discoverability of that content through relevant links. Users shouldn’t have to understand internal structures to get to that content.

The aim is to provide users with a way to find related content about a particular topic via a single, unified browsing structure.

OKR’s & KPI’s

As with any improvements you’re aiming to make to your site, ensure you have both OKRs and KPI’s.

An OKR could be increase # of pages per session by 10% by providing relevant, supporting content to users. KPI’s could be metrics such as average session length, average pages per session and/or bounce rate.

Whatever they are, be sure to agree on them with stakeholders prior to starting the project. This aligns everyone on the project and means that you can iterate on the taxonomy and aim for outcomes over outputs.

Content audit

First of all, you need to perform a comprehensive content audit of the entire site. This includes the following:

• what content exists
• is it redundant, obsolete, trivial
• is it up-to-date
• when it was last updated
• who owns the content (marketing, CEO, UX writer etc)
• who has editing rights to it
• page analytics
• readability scores (e.g. hemmingway)
• Any usability problems

Note: You may find content that doesn’t need to exist any more; for this you need to have a strategy around removing this content; whether that’s finding a new home for content, redirecting entire pages or removing navigation from around.

Creating the initial tags and taxonomies

Focus on labels which describes what content is ‘about’, rather than who it’s for, or what format it’s in. If you work on a large site or product, it may be wise to do a cross-section of the content first; test the labels (see below) and carry on once the labels are validated.

• label the content and record it in a database (spreadsheet or airtable is great) with each page or section of content
• use terms that reflect users’ language and mental models (this may require collaborating with stakeholders/subject matter experts)
• refrain from using internal names and descriptions
• add a hierarchy to tags to create a structured  taxonomy (e.g. Jobs > Volunteer jobs, Jobs > Paid jobs)
• Where it makes sense re-use previous labels

Testing the taxonomy

Run open and closed card-sorting with users to validate initial tagging. The taxonomy can then be  refined based on findings.

If you have a large amount of content run the card-sorting on a proportion, then continue labelling. Loop this process until all the content is labelled. Once all of the content is labelled, run a tree-test to validate the entire taxonomy.

Ongoing

• Try not to introduce any new labels for one year. All new content should fit within the existing label structure.
• All content contributors should use the one single list of user-centered tags
• After one year, any new tags need to be validated