Little Language Models

Think of a shoe.

What type of shoe did you think of?

My dad, who owns a tennis store and has played tennis for the last 30 years, would probably think of something like Asics sneakers.

If 10 readers responded, I’d assume your answers would differ based on different factors: your style, gender, what hemisphere you’re in (boots versus Birkenstocks), and your hobbies.

When you’re looking at a pair of shoes online, brands suggest similar shoe options. Considering the various factors above, how do brands know what to show you when their inventories include millions of products?

That is done through back-end content, the underlying architecture of a platform, which makes filtering possible and simplifies product discovery.

What is back-end information architecture?

Back-end information architecture (also known as content modeling) is the practice of structuring the technical content structure of a platform, usually in a diagram that depicts the various content types and their interrelationships within the platform, to facilitate discovery and increase content scalability.

When most people hear content strategy, they think of the circle on the left, but back-end content strategy is a powerful tool for customer experience transformations.

Investing in a well-designed back-end information architecture/content model offers several benefits:

• reduces the time it takes users to find related products or services
• enhances content visibility
• facilitates the discovery of new products
• reduces the time it takes to publish or update products
• increases content scalability by establishing modular content models covering different content and user types
• establishes shared understanding across disciplines and stakeholders of content type relationships
• facilitates effective planning for stages like research, UX wireframes, visual design, technical scoping, and copywriting

Who is back-end information architecture for?

• digital experiences with multiple steps, user types, and product offerings
• large, complex sites, apps, directories, internal platforms, plugins
• content-heavy businesses (enterprise, libraries, insurance, finance, educational platforms)
• high-growth startups and companies that want to scale

I’ve worked on some projects that didn’t require presenting the content model to client stakeholders. Some projects don’t even need a content model. Consider a self-published author’s website, designed to give readers a way to purchase the newest book. There’s no need to show the author how the back-end will function or make it complicated.

A rule of thumb: the larger and more complex the site, the higher the likelihood of someone misinterpreting their required involvement in content management.

Besides public-facing benefits of better findability and improved user experience, a content model helps establish a shared understanding across internal disciplines and stakeholders of content type relationships, ensuring that team members are informed of their responsibilities and required content management contributions.

What’s the format of back-end information architecture work?

The format of back-end information architecture work is usually a hierarchical diagram in Miro or Figma, intended for internal use only, depicting the various content types and their interrelationships within a platform:

• Post types
• Taxonomies
• Relationships between post types and taxonomies
  • Hierarchical relationships, if there are any
• Structure for each post type, including meta elements, taxonomies the post type can be mapped to, and design specs

I’ve created a simple content modeling template that you can download in Figma.

I’ve created tens of modular content models for clients and partnered with PM and engineering teams to guide implementation. I’d love to answer any questions you may have about the content model template.

Common back-end information architecture problems

Common pitfalls in back-end IA or content modeling differ from those related to user-facing navigation information architecture. They include:

• Lack of clarity on dependencies or future project phases affected by content modeling
• Failing to determine information needs
• No legend
• Not differentiating between automated and manual content
• Failing to acknowledge integrations

If you’d like a deeper dive into content modeling and common IA mistakes, check out the beginner-friendly guide, 16 common information architecture problems. The guide is packed with real-life examples and library scenarios to help you identify—and address—potential disruptions early on.