How to discover your content’s hidden mental model
How do you kickstart a content modeling exercise? All you gotta do is: figure out scope, agree on words, unearth your content's hidden mental model, and decide where to stop. Let's find out how...
In the first part of this guide, we learned about what content modeling is, and why content modeling (and structured content in general) is so valuable. This guide kicks off Part 2 – where we learn how to model content.
Before we dig into the specifics of how to do this, we need to understand how mental models work (well, at least a bit). They’re the ideas, labels, and archetypes that shape the way we relate things in our minds: collectively, as individuals, and in data.
Content modeling is really context-dependent. This means that the model you create for one project may have little, or no utility for the next one – even if they share the same domain.
With this in mind, we figured the best way to communicate the particularities of content modeling is to go through one from the perspective of a use case. That way we’ll be able to figure out some of the trickier sides of content modeling that can be hard to understand in the abstract.
With this test subject, we can sketch out priorities, talk about words, and offer tips for making sense out of all that stuff they've got.
We have a soft spot for sweet things... so we chose a candy manufacturer!
CandiCorp manufactures the world’s tastiest and most interesting treats. They are headquartered in sunny Skarsvåg, Norway, and have regional offices in Belgium, Singapore, and the United States.
CandiCorp’s primary revenue stream is through wholesale sales, which they expedite through a network of global distribution partners. Partners receive an annual product catalog of sweet deals every year, which is delivered in print.
Why they’re making changes
CandiCorp management have identified a range of inefficiencies in the way their content is managed. They’re also piloting a direct-to-consumer (DTC) candy subscription box program for customers in the European Union. The time is ripe for them to re-evaluate the way their content is structured, so that they can:
- Setup a new subscription program.
- Manage print and web product content from a single location.
- Better management of organisational staff and contact information.
It’s important to have a sense of how far you want to go in your modeling exercise. One of the best ways to do this is to have a pretty concrete sense of what you want to achieve with it. So if you can, create a high-level reference that includes things like:
- The problems you need to solve for the organisation
- Things you can improve for your audience
- A sense of the places where you can bring more value: either internal or user-centered
- The amount of time and resource you want to spend on the process itself, and the downstream work that it will generate. (If you're thinking about skipping the modeling part, we think you should reconsider, and here's why).
While it can be tempting to try for a content model that’s fit for all purposes. We think it’s best to lean on the side of pragmatism and focus your efforts on the specific goals you want to achieve, rather than attempt to model “reality” one to one.
Think of your content model as a map that you can use to navigate a landscape; containing only the information you need and nothing more.
Card sorting “reality”: a scene from the film “Synecdoche, New York” (2008) where playwright protagonist Caden Cotard, under an unwavering commitment to representing reality in his show, endlessly elaborates on capturing the details of the lived experience from the city in which he lives. Spoiler: it doesn’t end the way he planned.
There are established methodologies you can follow to tie organizational goals to content design. We would recommend looking into The Core Model as a way to tie strategic goals to user tasks. While this doesn’t address content modelling per se, it’s a useful way to start thinking about what content you need to connect user tasks and strategic goals. We’ll return back to the Core Model later in this chapter.
CandiCorp’s project scope:
- DTC subscription program for EU residents
- A searchable online product catalog
- Manage product, organisation, and marketing data from one location
Nice to have:
- A dynamic online staff directory
- The ability to generate an organisational chart from staff records
- A place for pets in the organisational directory
- More efficient content workflows
- The ability to publish time-sensitive content related to new subscription issues
- Easy sign on to the new subscription program
- The ability to define subscription size and frequency
- A rich, searchable online catalog
- More useful and valuable product information
- A small working group to see the modeling process through
- Representation from key departments (marketing, exec, development)
- Audience insights (in the form of user research)
Gather up a working group of contributors. Their job will be to represent the varied needs of your organization and audience, and implement the new model in code. The amount of people in your working group depends on the scope of your project.
While it’s important to canvas all the ideas and expertise that could benefit the goals of your project, don’t make your working group so large that you can’t get traction on your tasks.
To get the most out of your team’s time, set up a great environment for working together. Agree to put away distractions such as laptops, cell-phones, or if you do it remotely, mute all outside communication applications. Remember to take breaks and stay hydrated and keep the blood sugars on a healthy level.
For more advice on how to make for great group sessions, check out the Design Sprint book by Jake Knapp and John Zeratsky.
Our crafty confectioners assemble a crack squad of valued contributors from across the organisation – the Content Model Working Group (CMWG). While they’re a big company, they decided to keep the working group small, yet representative of all the main stakeholders:
- Head of marketing
- Content marketer
- User researcher (external consultant)
- Subject matter expert
- A representative from the executive level
The idea here is to make a high-level map of the ideas and materials that you’re working with. You need a collective mental model in order to reason about your content in effective ways, and have it meet the needs of all those who author and consume it. Parts of this process are sometimes called “sense-making”, or “domain discovery”.
How it works:
- Agree on words
- Identify where business and user needs cross over
- Establish reasonable domain boundaries
- Document what you know
Exactly how you go about this process is up to you. Here are some factors that may influence how you tackle it:
- Whether or not your team is co-located
- How tech-focused you all are
- Whether or not you want to actually use technology to facilitate the process
Processes like card sorting, software like Whimsical or Miro, and spreadsheets can be helpful for what’s to come, but you don’t have to use any one thing in particular. The main idea is to work in a collaborative manner and document your process and outcomes in a medium that’s easy work with. Whiteboards, paper forms, and sticky notes work just fine!
Words mean different things to different people in different situations. So, it’s important to ensure that you’re all talking about the same thing when you use a certain word or term. If your team comes from different perspectives within the organisation (which we hope they do!) it may take a minute to find common ground.
You can record the important terms in a document, or if that’s more than you need – simply encourage an environment where everyone can speak up if confused. It’s all about finding common ground for what comes later.
The study of ethnography has developed a few interesting methods for uncovering the domain terms used by social groups.
One of these is called Freelisting, where you attempt to uncover all the examples of a ‘thing’ that people in a domain or social group operate with. Once you have all the terms you can figure out which ones carry the most weight based on their frequency and distribution. It’s an effective way to parse a large amount of data in a few minutes without the need for special tools or expert facilitators.
For example, you could ask the employees of CandiCorp to enumerate what types of candy they know about, or to list all the product bundles they are aware of. The results can be surprising (or not).
It's really easy to overlook this part. You can provide a lot of value to your audience simply by using common words and terms that they already know and use. These terms are really useful for your content model, but can serve double duty for audience-facing material such as navigation labels, taxonomies, and taxonomy terms.
By using transparent and commonplace terms you’re doing everybody a favor. When it comes to user interaction, a label is a promise – anything you can do to make them live up to that promise is a good thing.
Any broken promise, large or small, chips away at trust and credibility. The words in a link label make a strong suggestion about the page that is being linked to. The destination page should fulfill what the anchor text promises.
Kara Pernice – Nielsen Norman Group
Lost in translation: the cost of not agreeing on terms
In 1998, NASA launched the Mars Climate Orbiter into space. This 638-kilogram robotic space probe set out to study the climate and atmosphere on Mars. 10 months later the probe burned and broke into pieces in the red planet’s atmosphere (source).
Investigations into the crash revealed that of the two main teams working on the project: one used imperial units of measure, while the used metric. The result was a spacecraft lost-in-translation. These kinds of problems can be easily remedied by taking time out to agree on words early on.
All content fulfills business or user needs of some kind. But the best value exchange lies in content that provides for both.
If you make content that focuses solely on the business, your user may not find any value in it, and it can perform poorly. Most of CandyCorp’s customers will not care deeply about their company news section.
On the flip side, a business that produces content solely for user interest may not see a sustainable return on their content investment. The trick is to provide structured content opportunities that allow for a fluid exchange of value between those who provide the content, and those who consume it.
Does this guide achieve that objective? We’d like you to learn to become a better content modeler no matter what tools you use. But... we’d also like for you to have a favorable impression of Sanity, and at least consider it as a worthwhile solution for structured content management! 😉
The Core Model is a great way to find the sweet spot between both sets of needs. It does include some focus on page-based concerns, but the principles are translatable to content modeling.
The Core Model: a handy technique for bridging business and user needs
In 2007 an information architect by the name of Are Halland presented his design framework for findability, prioritization, and value that he called the Core Model.
The model’s premise is to build web pages and other digital products from the inside (core) out by making the unification of business and user needs the primary concern of each page and product instance.
"The core model is first and foremost a thinking tool. It helps the content strategist identify the most important pages on the site. It helps the UX designer identify which modules she needs on a page. It helps the graphic designer know which are the most important elements to emphasize in the design. It helps include clients or stakeholders who are less web-savvy in your project strategy. It helps the copywriters and editors leave silo thinking behind and create better content.
...by the end of the workshop, the group will have a common understanding of user needs, business goals, and how different pages should be connected."
– Ida Aalen
If you’re finding that it’s hard to get a high-level sense of what your organisation and users really need, you may want to give the core model a try.
This is a little bit like defining scope, but applied to your subject domain.
Everything we do is in relationship to something else: be it chemical, ecological, cultural, or otherwise. It’s also the case with content modeling, domain mapping can be a bit like mapping the world: one thing relates to another, then another, ad infinitum. So, if you don’t want to end up like Pepe Silvia, you'll need to establish some reasonable boundaries for your domain.
A small boutique candy maker may exist in the same domain as CandiCorp, but they probably don’t need to do as many things with their content. It’s all about finding the sweet spot (pun intended) where your model captures the relationships that are most important to your needs, and ignores (at least for now) the edges of your domain where immediate value is not obvious.
CandiCorp’s new DTC subscription concept presents them with an interesting conundrum:
How to manage subscription customers and payments?
Up until now, their immediate customer base has been partner wholesalers. But they’re a relatively small group of customers compared to the numbers of people that may sign up to the subscription service. Distribution partners also operate under a pre-approved line of credit whereas new subscribers will need to initiate their own signup and make regular automated payments via credit card.
Ontologically speaking, partners and subscribers are both a type of “person”. While they could both fall under a single person content type – there’s enough difference between the two to justify keeping them separated in the model. They know they can unify these types later if they want to.
While partners and subscribers are both “persons” with a lot in common, the differences between them indicate that they’re two distinct groups. Remember, the content model’s job isn't to represent reality 1:1, but to make it feasible to achieve some specific goals.
On the payments side: they could use Sanity to build out all of the required functionality of a fully-fledged system, but they’re wary of investing so much into a pilot program – after all, they’ve never sold direct to consumer, and while marketing is confident that things are gonna work out great, there’s no guarantee of success.
In the end, they opt for a minimum-viable pilot:
- Treat partners and subscribers as separate entities so they can get to market faster.
- Rely on a 3rd-party SaaS solution for storing and managing subscriber data and subscription payment processing.
While this greatly reduces the boundaries of their domain for phase one, they’re also mindful of the fact that they’d like to eventually in-house everything operations if things go well. So they commit to setting up their data model so that they can extend it into a more fully-featured system later on.
By this stage of the process, you should have a pretty good mental model of what the most important content types are, and how they (generally) relate to one another in order to fulfill you and your audience’s needs.
If you’ve been card sorting you may have collated groups of items that can represent a particular content type or category. Perhaps you diagrammed your way through the process, or have spreadsheets and pages of things?
Now it's time to assemble what you have and crystallize that thinking into a documentation that includes an indication of:
- Your main content types
- The attributes you need to relate them to one another
A helpful way to document this is to add short and helpful descriptions for the fields where your editors work, as well as adding validation where needed. That being said, it may also be useful to keep a document with more elaborate documentation so it’s possible to refer back to the reasoning and intention when you want to iterate and expand your content model.
This doesn’t have to be exhaustive and time-consuming, because you’ll be adding plenty of detail in the not too distant future. It’s a map of sorts that you can work with, but the most important 'map' at this stage is the mental one that you hold in your collective hivemind.
The working group's round of sense-making and discovery emerged with consensus in the following areas:
- They need a subscription type to connect other things to
- Subscriptions need customers
- This customer group is new and has unique needs, so it’s a new content type
- Subscriptions will change throughout the year – they need the ability to make subscription issues
- Each issue needs new combinations of products to make the subscriber experience interesting
- Products still need to be categorized (they have this for their printed catalog already but need more use from it)
- The catalog lists products
- The new website will have collateral in the form of time-sensitive articles and evergreen landing pages for SEO and sales enablement
They drafted a basic schematic to validate the things they agreed upon, and include some items that needed relating later on.
Questions for the future:
While the group agreed that these foundations were pretty bulletproof, they had LOTS of questions like:
- Should subscribers have the ability to choose the size and frequency of their subscription? If so, how will that variance be reflected in cost?
- Should we include a newsletter with each subscription issue? And should this have crossover with the website articles?
- Is the current product categorization approach enough to fulfill the needs of an interactive online catalog?
- Does the online catalog need a linear presentation (like the print version) or will search and sort functionalities suffice?
- Should articles make their way into print and/or web catalog?
- If both: how will long-form content accommodate writing for the non-linear hypertext style of consuming content on the web (with links connecting things) and the linear reality of print publication.
- Should articles have authors? And do we need a new content type to support that?
- How can we include products and subscription materials in landing pages while avoiding unnecessary duplication of content, or presenting outdated information?
- How can we support staff and departmental hierarchy to maintain a living organisational chart and directory?
We came to understand the importance of understanding the scope before modeling content. We worked through ways to make sense of our content and ambitions, and relate them to the task at hand. While we were there we considered how our audience might make sense of our content, and whether our labeling and frames of reference would fulfill their needs. Finally, we reached a consensus on the essential pillars we need to fulfill the project’s ambitions and documented a whole bunch of questions that will kickstart the model’s next steps.
Next up we’ll build these pillars in Sanity, and begin connecting them to one another in order to bring durable workflows and exceptional content to CandiCorp and their awesome customers.
Before you go: don’t forget feedback!
It’s a good idea to reach out for feedback before you start implementing your model in code. This can be especially helpful if your working group is small and the project scope is large.
Perhaps you couldn’t afford to budget time for a user researcher to attend the entire modeling exercise? Now’s a good time to share the core ideas that you came away with.
Input from others can reveal important pieces of the puzzle, and it’s still pretty cheap to change your mental model, but once your dev and content operations teams get underway things get more expensive to change.