Lenses on Information Architecture

Cross-Cutting Topics

Published in

EightShapes

10 min readAug 12, 2022

In any given domain of information, some topics that cut across every other one. Hence: a cross-cutting topic. In medicine, for example, you have cardiology, gastroenterology, ophthalmology, and pediatrics, among others. Pediatric medicine can also cut across all of them: you can talk about the health of the heart, the gut, or the eye in adults or in children.

But we usually see pediatrics as a topic alongside the others, as if cardiology and pediatrics are both at the same level. And in some contexts, a straight list of topics — no cross-cutters — is most helpful. Physicians must choose to specialize, so they might choose general cardiology or general pediatrics.

This is the state I usually find a set of topics in: a long, straight list that does not acknowledge that some topics have evolved such that they cut across the others. I’ve seen this in medicine, real estate, employment services, digital storage, and many other domains.

In every domain, at least one topic that has been treated the same as any other in fact serves as a lens through which to see the other topics.

What’s remarkable isn’t so much that some topics evolve to become cross-cutting. It’s not even surprising that this happens consistently across different fields and industries. Instead, what I find remarkable is the organization’s dogged adherence to a flat structure. The expectation, I suppose, is that the information has always been categorized in this way, so this is how it will always be categorized.

The Emergence of Cross-Cutting Topics

How does a topic become cross-cutting? I’m doing a bit of reverse-engineering here, looking at the attitude of the domain experts toward the domain’s topics.

Increased Emphasis

Over time, one or two aspects of a domain might become more interesting or relevant, due to changes in circumstances or increased attention on it. I’ve seen this time and again with the concept of “sustainability”. Once thought to be a single plank in the platform, organizations have come to realize that sustainability is a concept that pervades every aspect of their business. Accessibility is not there yet (frustratingly) — organizations treat accessibility as a separate topic next to all the other aspects of their business, but should instead think of it as a lens through which to look at everything else.

Over time, some topics become more important, perhaps because society has placed more value on them. These topics can become cross-cutting — that is, a lens through which to see all the other topics.

Organizations focused on offering a set of services sometimes treat “technology” as a distinct category. You can almost imagine their marketing materials from the 1990s cheerfully offering technology services alongside their other offerings. Three decades later, organizations acknowledge (hopefully) that technology pervades every service they offer.

Emergent Matrix

Sometimes it’s more than one or two topics that become cross-cutting. In these cases, the domain has become so nuanced that topics can be organized into a matrix, and that content is only interesting at the intersections of topics.

An obvious example of this comes from our work with large multilateral organizations. Historically, the content was siloed by topic, by content type, and by geographical region. Analysis showed that although it may be useful to have pages for broad topics like “environment” and “infrastructure”, these became more meaningful when intersected with geographical regions.

In some domains, the main topics have historically been treated as siblings when in fact they represent different perspectives. Subjecting these domains to more scrutiny, in the attempt to digitize their content, it becomes clear that these topics must be distinguished.

Our work in physician training also shows that the industry generally relies on lists of topics treated homogeneously. So, Women’s Health is lumped in with Cardiology. But topics that deal with demographic groups (like “women’s health”) or medically relevant populations (like “immunocompromised”) or new technologies (like “imaging”) have relevance across more traditional medical topics. I predict that as medicine embraces a more holistic view of human health — the body’s systems don’t act in isolation — we’re going to see much more matrixed content.

Validating Cross-Cutting Topics

Looking at a domain with fresh eyes reveals that a formerly flat list of topics has some cross-cutting topics. But these are only hypotheses, and they need to be tested. Like any information architecture research, it’s difficult to arrive at a clear direction. Instead of seeking definitive direction through research, I seek feedback on the concept using a few common techniques.

First, I can do some validation without any testing whatsoever. An information architecture is only as good as its ability to successfully frame the content. So, the first thinking tool I use for this is stress-testing. By stress-testing I mean using the categorization scheme to classify real content. I look at various combinations of topics and evaluate whether there’s meaningful content at their intersections. In short, I’m asking myself, “How much content do I have at the intersection of Topic A and Topic B?”

Another way to stress-test the concept is to use a content inventory. Like most of my work product, the content inventory is in a spreadsheet. In this case, each row represents a distinct piece of content. I add a couple columns, one to indicate the main topic and one to capture any cross-cutting topics that might be represented in the content.

Cross-cutting topics can manifest in at least two ways. We can use cross-cutting topics to “triangulate” the location of content in the structure. Alternatively, cross-cutting topics might inject interesting content into a main topic page.

Each of these stress-testing approaches addresses different ways in which a cross-cutting topic may manifest relative to a main topic. The first approach asks whether there is stand-alone content that may be classified by both a main topic and a cross-cutting topic. The second approach asks whether some of the content on a given page might be classified with a cross-cutting topic.

To be clear, one or two negative results from stress testing doesn’t invalidate the entire concept. Cross-cutting topics are interesting because they highlight meaningful and important aspects across a domain. Reality is messy, however, and those aspects may not be consistently relevant across the domain.

Tree-testing is a technique that gives participants tasks to locate content in a tree of categories. They navigate through the hierarchy until they find the category they think the content is located in. For domains with cross-cutting topics, I set up the tree to include both the main topics and the cross-cutting topics. This is tricky, because it may not be an accurate representation of how you want to set up the navigation, but it is an accurate model of the underlying information structure.

When testing cross-cutting topics, you may need to fake the navigation to expose both the main topics and the cross-cutting topics.

The success of this approach depends on writing real-world tasks that involve content at the intersection of two topics:

For medicine, it might be looking for a cardiology case study involving an immunocompromised patient.
In domains heavily reliant on geography, the task could involve seeking out transportation infrastructure projects in sub-Saharan Africa, for example.
For a business that focuses on providing services, I like to ask users to find information about technology solutions that support the services offered.

Tasks like these allow you to discern which aspect of the domain users prioritize when seeking information.

The results might show that users favor finding this content with one approach over another. You might observe that most people tackle the “transportation infrastructure projects in sub-Saharan Africa” task by clicking into the region menu. In this case, the geography looms large in your users’ imaginations, an insight that can help you determine how to treat cross-cutting topics in the menu design.

On the other hand, if things are evenly split — some users clicking on the subject matter and some clicking on the region — you can surmise that the mental model isn’t stable across users. The tree-testing tool I use — Optimal Workshop’s Treejack — allows you to dig into the path that participants followed. You might see that users went straight for one category or the other (high directness) or, they bounced between the two before settling on one (low directness). All these hints suggest how you might design the menus to ensure visitors see the keywords that resonate with them.

Designing for Cross-Cutting Topics

With an information architecture involving cross-cutting topics, there are a several important design decisions. The answers to these questions depend entirely on the design principles you establish for yourself. Those principles, ideally, are grounded in a thorough understanding of the domain’s content, the organization’s objectives, and, of course, your understanding of the audience’s expectations.

How does a cross-cutting topic appear on a main topic page?

There’s no particular order to these decisions, but I do like to start with how the cross-cutting topic is integrated into the primary “back bone” of the information architecture. If the spine of the IA deals with a set of subjects, the cross-cutting topic is a lens through which to look at each subject. For example, if Transportation Infrastructure Projects has a page, how are geographical regions (which are cross-cutting topics) represented on the page?

A main topic page can include sections for short content about each of the cross-cutting topics.

Each region might have a distinct section on the page. The breadth and depth of those sections depend on the purpose of the site and the available content. Maybe the organization just has a couple sentences they want to say about Cardiology and Women’s Health. Maybe they have an entire library of content.

Do cross-cutting topics need dedicated pages?

Whether or not a cross-cutting topic appears on a main topic page, it could have its own stand-alone dedicated page. A page on Women’s Health, for example, might talk about the organization’s overall philosophy and approach, irrespective of how it connects to the main topics.

In turn, the content “hooks” on the main topic pages can link to pages with more content about the cross-cutting topic.

A cross-cutting topic might deserve its own page if there’s a stand-alone story to tell about it. It might also be the kind of topic that users come looking for, and having a dedicated page gives it some weight. Finally, a cross-cutting topic could offer a good way to navigate the main topics, highlighting those that are further illuminated by the cross-cutting topic.

Finally, the cross-cutting topic page can link back to the main topic page.

Do topic intersections need dedicated pages?

A third decision is whether the site needs pages that are dedicated to the combination of a main topic and a cross-cutting topic. For example, instead of (or in addition to) having a sliver of content on the Cardiology page about Women’s Health in Cardiology, there might be a page dedicated to all the content about Cardiology + Women’s Health. I call these “intersection pages,” because they expose content at the intersection of two topics. These pages offer niche content which might differentiate the site from other organizations’ sites. The intersection pages create additional space in the structure in case a blurb on the main topic page doesn’t cut it.

In addition (or instead) of stand-alone cross-cutting topic pages, the structure can include intersection pages. Pages with content that sit at the intersection of the main topic and the cross-cutting topic.

One aspect driving this decision is whether there’s sufficient content and/or interest in these intersection pages. Every new page represents an additional level of effort for managing and curation. The interest or available content for these intersections may be uneven. (Example: Cardiology + Women’s Health may be rich with possibility whereas Ophthalmology + Women’s Health is probably not.)

Finally, given the breadth of content on the site, it may be best to automate the content on these pages. That is, since there may be many possible intersections, it may not be realistic to manage all those pages independently. Instead, the underlying content management system assembles the content by finding everything that’s tagged with both categories. This represents a different kind of effort, designing the rules to populate these pages in various circumstances.

How do cross-cutting topics appear in the navigation?

Save menu design for last. It helps to understand what the structure will be — whether cross-cutting topics have dedicated pages or not, for example — before you design the wayfinding mechanisms. Cross-cutting topics are tricky for menus because they do appear across the main topics. If each main topic + cross-cutting topic intersection had a dedicated page, the menu might appear littered with redundant labels.

In the navigation, it may be tempting to put a link to each cross-cutting category under each main topic. This approach, though, creates a lot of noise and doesn’t paint a concise picture of what’s available.

Remember that the navigation menu does not need to be comprehensive. It better serves users by depicting the range of things available and highlighting useful starting points. If the cross-cutting topics are good starting points, create a list of them that implies they are orthogonal to the main topics. If it’s instead important to show that your main topics have these cross-cutting aspects, you can point to them in a more oblique way.