Information architecture on government websites is a governance problem, not a design problem

Information architecture on government websites keeps breaking in the same way. Users cannot find what they are looking for. Menus grow without structure. Content gets buried three clicks down under a heading that makes sense to the agency but not to the person visiting the site.

The standard response is to redesign the navigation. Run a card sort. Do a tree test. Build a new sitemap. Iterate.

The navigation improves. Then, within 18 months, it starts filling back up.

I have seen this pattern across multiple government digital projects. The structure degrades not because the original design was wrong, but because the organisation lacks any agreed mechanism for deciding what goes where and who has authority to change it.

That is not a design problem. It is a governance problem.

What information architecture actually requires

Information architecture — the structure and labelling of content on a website — is often framed as a UX discipline. It involves research, testing, and design. That framing is not wrong. But it misses the organisational conditions that determine whether an IA holds up over time.

Good IA requires three things that have nothing to do with the original design: a decision authority who can say no to new content requests that do not fit the structure; a clear taxonomy that teams across the organisation agree to use; and a review cycle that checks whether the structure still matches how users look for things.

In most government agencies, none of these three things exist in a durable, documented form. Content gets added wherever there is space. New programs create new sections without reference to existing ones. The navigation is maintained by whoever has publishing access that week.

What the research says about why structures fail

Information architecture — the structure and labelling of content on a website — is often framed as a UX discipline. That framing is not wrong. But it misses what determines whether the structure holds over time.

Kristina Halvorson's foundational argument in Content Strategy for the Web is that structure and governance are distinct disciplines that require separate investment. IA defines where content lives — the categories, the hierarchy, the labels. Content strategy defines the people, processes and policies that determine whether content stays there, gets updated when things change, and gets removed when it no longer serves a purpose. Most government IA projects commission the first without the second. The sitemap is delivered, the navigation is built, and the project is considered done. What is not delivered is any mechanism for maintaining the decisions that produced the structure.

Patrick Lambe's research into why taxonomies fail in organisations reaches the same conclusion from a different direction. Having studied taxonomy projects across sectors, Lambe found that classification errors are rarely what causes a taxonomy to break down. The original categories are usually sound. What fails is ownership. Nobody is assigned to make calls about where new content fits, resolve edge cases, or update the structure when the organisation changes. Without that, even a well-designed taxonomy quietly absorbs whatever it is given — until the structure no longer reflects anything meaningful.

Both arguments point to the same gap. An information architecture can be methodologically rigorous and still degrade within 18 months if no one in the organisation has a standing brief to protect it.

Why IA redesigns fail

Most IA redesign projects treat the structure as the deliverable. The research is done, the new navigation is built, the card sort data supports it, and the project is considered complete.

What the project does not deliver is the governance infrastructure that would keep the structure intact.

No content taxonomy linked to content types. No process for requesting a new section. No one with a brief to say "this does not fit here."

The Australian Government Style Manual covers how to organise information for users. It does not cover how to govern the structural decisions that determine where content gets put over time. That gap sits between the content design work and the organisational systems that support it.

What a governance-backed IA looks like

A durable information architecture for a government website needs three things.

A documented taxonomy. Not a list of page titles — a defined set of categories and the rules for assigning content to them. The taxonomy should be small enough to use consistently and specific enough to resolve ambiguity.

A named decision authority. Someone with authority to approve new sections, approve exceptions, and say no to content requests that do not fit. In practice this is usually the digital team lead, but it needs to be explicit rather than assumed.

A structural review cycle. Every 12 to 18 months, check whether the top-level structure still matches how users search for and navigate to your content. WCAG 2.2 Success Criterion 2.4.6 requires that headings and labels describe the topic or purpose. Getting this right at the IA level — not just the page level — requires the same structural rigour.

The practical question

If your government website's navigation structure is degrading, ask: who in your organisation currently has authority to decide where new content goes?

If the answer is "it depends" or "whoever has publishing access", you do not have an IA problem. You have a governance gap. The structure will keep degrading until the accountability question is answered.

Redesigning the navigation is the visible work. Building the governance that protects it is what makes the redesign last.

Photo by Piotr Wilk on Unsplash.