Thinking beyond the limits — UX expert Michael Atherton on using content models in content strategy

Why use content models when working on a content project? How can they help you optimize your content strategy across various channels? Michael Atherton, UX design instructor at General Assembly in London, tells all.

Content modelling is one of the most overlooked techniques that you can employ to structure your content before making strategic decisions about when and where to use that content.

A content model gives you a visual overview of the various content types in your inventory and shows how these types are or are not related to each another.

Making a content model forces you to think about how the building blocks of your content offering are connected, paving the way for further IA work such as the creation of site maps, page structure, navigation flow and the like.


How can content strategists and information architects both benefit from using content modelling to structure their content?

Content modelling is a great technique for structuring content based on the meaning of that content, rather than on the classification of document types.

As humans, we tend to think about real-world things and the real-world relationships that connect them, rather than about documents about those things.

By defining those things and relationships, people are able to navigate content more fluidly and learn how a subject hangs together based on the connections within that subject and across different subjects.


Visually, a content model looks a lot like mind map or a site map, but it’s actually not either of those. What differentiates a content model from them?

It does look like a mind map. Your boxes are the concepts or things within a subject (so for ‘Food’, it might be 'recipe', 'ingredient', 'chef', 'technique', etc.) and the connecting arrows describe how these concepts are connected.

The real power is that those relationships are also defined – a chef 'created' a specific recipe – say, Eggs Benedict – and hollandaise sauce 'is an ingredient in' Eggs Benedict. So unlike a site map where the relationships are implicit, based on similarity, in a content model the relationships are explicitly stated.


Example of a content model for a cooking website, using recipes as the main content type. Note the strong connection between ‘Recipe’ and ‘Ingredient’: a completed recipe could become an ingredient in another recipe, such as hollandaise sauce in Eggs Benedict.

Content modelling is not commonly practised among IAs because it’s perceived as a hard thing to do. Where does that bias come from?

The basic process of creating a content model is simple. Taking it to the web becomes a bit more complex because many content management tools impose basic hierarchical structures.

A content model translates more readily to a relational database schema, which means IAs have to get friendly with database architects and use CMSs that can support rich hypertext relationships between content.

It's getting easier, though. Tools such as Drupal already support custom schemas for content.


It’s clearly a good skill to master when dealing with larger content structures. Is it also suitable for small content projects?

Absolutely. If you’re in the business of providing content, that content needs structure.

An underlying principle is to think of your content offering as contributing to a subject domain. Got a movie showtimes site? You’re in the domain of cinema, with films, actors, awards, genres as your ‘things’. Concert tickets? You're in the music business, specifically live music, with artists, lineups, venues, and set lists.

Even if I was doing a site for a small chain of restaurants, structured content helps. People probably care about the subjects of food and fine dining more than they care about my restaurant. Hence, we can talk about the cuisines, the dishes, the locally sourced ingredients, recommended wines, our fantastic chefs, and the food we serve that’s suitable for vegans or coeliacs.

We can offer all of these interesting things as ways into our content and as the means to connect everything together.


Many people, when they think of IA, think ‘website’ almost automatically, but content modelling goes beyond that. Why is it such good practice to look at the broader, multichannel picture?

We're publishing content to the most interconnected knowledge repository in history – the internet – and we should always consider the implications of that social contract. Think beyond pages, beyond the walls of your website, and even beyond the limits of your content offering.

Think instead about how your content contributes to the body of knowledge that the internet has to offer about a subject, and thus where you fit within that wider mental model. What take on that subject can only you offer?

Then make that content granular and atomic enough to map to the things that users think about – and search for. Give them an easy way to pinpoint the specific concept they want to reference, and they’ll reward you with links and traffic from search engines.

Connect your concepts at the data level, not just the page level, so that the semantic meaning is inherent across all device representations, and readable not only by humans but by robots, by services like the Google Knowledge Graph and by any number of semantic products to come.

Accept that you can’t cover everything about a subject, nor is anyone asking you to. As users, we cherry-pick information from across the whole web, so find where your offering touches that of others and link as much as you can, stitching your content ever tighter into the fabric of the web.


A solid content model can not only help strategists and IAs, but SEO specialists as well. Why is that?

When a content model does get represented as a website, it often results in a series of single pages each dedicated to a single instance of a thing.

As you might find with Wikipedia, your own site could end up with a page for the Mountain Lion, the Sumatran Tiger and the Spectacled Bear (oh my!). This is great for SEO, as search engines love specific content focused on a search term. And those concepts mapped out in the content model typically map to the level of granularity we find in search queries.

SEO also loves link density. Publishing many discrete and focused pages, each at a unique URL, makes your content easier to link to, to cite and to tweet about (again, look at Wikipedia's success there). This increases external link density – the number of third-party sites linking to your content. Also the rich hypertext linking between concepts within your own site boosts internal link density.


What are some of the broad-stroke lessons you can take away from a content model that has been done right?

A good content model demonstrates empathy for its subject domain.

Start by talking to subject-matter experts to understand the subject domain that your content occupies. What are the important concepts in that domain? How do they connect? Then talk to users. How is their understanding of the subject different? Which things are most important to them?

Aim for a best-fit model with enough authority to please the experts, but enough accessibility to be useful. Experts map the territory; users mark the points of interest.


Michael Atherton will be holding a workshop on modelling content domains at EuroIA in Brussels on 26 September 2014. The workshop will help beginners understand the benefits of structured content and will give them hands-on experience in making models. Follow Michael on Twitter: @MikeAtherton.


Are you looking to set up or optimize your content strategy? Please contact Pieter Vereertbrugghen for more information on how Cypres can help.