Trying to fix an organization’s content problems by installing a content management system (CMS) is like trying to save a marriage by booking a holiday. We know that a successful web project needs a content strategy—but when it comes to the CMS, we stop thinking strategically. Despite all the talk about user-centered design, we rarely consider the user experience of the editorial team—the people who implement the content strategy. We don’t design a CMS, we install it.

The problem: tools aren’t magic pixie dust

Any web project more complex than a blog requires custom CMS design work. It’s tempting to use familiar tools, and try to shoehorn content in—but we can’t select the appropriate tool until we’ve figured out the project’s specific needs.

WIREFRAMES ARE ASPIRATIONAL FANTASIES

As Karen McGrane says, it’s easy to sketch a faceted navigation on a wireframe. It’s more difficult to implement a CMS to power the implied taxonomy, and to commit to ongoing editorial maintenance over time. A wireframe without a corresponding content strategy and a realistic CMS design is a work of fantasy. A CMS that could realize one of these fantasy wireframes would need plenty of magic pixie dust. We need content strategy to help us decide which of our aspirations is feasible; CMS design is an essential part of that decision.

USE A DESIGN PROCESS TO SELECT AND CUSTOMIZE A CMS

Most of the time we select a CMS by popularity, cultural affiliation, or corporate edict—that is, without properly considering the content we’re supposed to be publishing. This is crazy. Instead, we should use a design process to select and customize a CMS, based on our content strategy and the editorial team’s needs. This article will show you how. But first, let’s take a step back: what exactly is a CMS?

A CMS is a bunch of features

We’ll define a CMS as a set of software tools that enable non-technical people to manage web content. There are a bazillion different CMS tools out there. They tend to be sold on their features—and boy, do they have a lot of them. Here’s a taste:

• content creation and editing,
• content delivery,
• taxonomy management,
• curation and page composition,
• editorial workflow, and
• …continued ad infinitum.

Now, I love me a good CMS feature. But features alone can’t solve strategic, editorial, or governance problems. Too often, CMS projects become solutioneering, or throwing technology at problems. So what should a CMS give us, apart from a bunch of features?

A CMS SHOULD ADD SEMANTIC RICHNESS

To get value from a CMS, think beyond editing web pages. As Jeff Croft argues:

…[content management] ought to include structuring, organizing, searching on, filtering, and easily modifying your content… [allow you to] quickly define new types of content… [and] facilitate establishing meaningful relationships between disparate pieces of content. It ought to make your content more useful simply by virtue of the content being in the system.

Web pages are where our content ends up, but using a tool that can only edit pages is like marking up document headings using the div element; it might look fine on the surface, but proper, semantic heading elements are more useful. So a CMS needs a rich content model that can generate semantic web pages.

A CMS SHOULD BE CUSTOMIZED FOR A PROJECT’S NEEDS

A product can’t fix content problems out of the box. Every CMS started out as a solution to a specific problem, later generalized to fit a wider range of problems. It’s an open secret that CMS tools need to be customized before they can be used on a real website. As D. Keith Robinson argues:

In truth most CMSs end up being custom, regardless of how they start out. From those that bill themselves as one-size fits all to the highly specialized systems which deal with specific industries or types of content. It’s just a matter of how much hacking you’ll need to do to get to what works for your people.

So a CMS is a bunch of powerful tools that add semantic richness to content, and that require customization to fit a specific project’s needs. Before we learn how to to apply them to a content strategy, let’s take a short history lesson.

Content strategy is shaking up the CMS industry

The rise of content strategy is dealing the content management industry a huge kick up the backside. In the web’s Wild West era, the CMS was run by the IT department—or sometimes a lone webmaster who knew HTML—so CMS choices were based on features, price, and cultural fit, rather than web or content strategy. It was the classic IT drill: selection committees, feature matrices, and business lunches with men wearing neckties.

The times they are a-changin’. According to Lisa Welchman, the web is now “the organization’s primary communications, sales, marketing, and transactional vehicle.” A CMS vendor’s target audience used to be the IT Director, and a successful outcome meant each department could easily update their content silo. Now, the target audience is an organization’s internal editorial infrastructure; and a successful outcome is a complex mix of achieving business objectives, implementing a content strategy, and crafting a user experience. The game just got more serious.

A process for selecting and customizing a CMS

We choose CMS tools for crazy reasons. See if you recognize any of these scenarios.

• Choosing a tool because someone you admire uses it—and expecting results like theirs—is like buying the type of guitar Jimi Hendrix played and hoping to fill Madison Square Garden next week. (Giveaway: “All the cool kids use ACME product.”)
• Choosing a tool based on your cultural attachment to it makes the project more about you than your client’s objectives and your users’ needs. Avoid holy wars. (Giveaway: “ACME product should be your next CMS.”)
• Choosing a tool because the IT department says you have to is like accepting an artistic commission while handcuffed; it’s possible to do good work, but you’re set up to fail. (Giveway: “The client requires ACME product.”)

It’s as if we’re considering every factor apart from the content. Let’s take the advice of Karen McGrane and Jeff Eaton, and “reframe the conversation”:

Shift the discussion about the CMS away from “features” and towards “task flow.”

So what does a grown-up CMS selection process look like? Here’s a workflow diagram:

The inputs are the content strategy, which is made up of substance, structure, workflow, and governance; our editorial resources, i.e., the editorial team’s ongoing time commitment; and our technical resources, made up of infrastructure (e.g., hardware) and our technical team’s time commitment. Using the complementary design processes of content modeling and task analysis, we’ll create a CMS selection & customization plan, that describes which tools to use, how we’ll customize them, and how we’ll maintain them over time. Let’s take each in turn.

The gold mine: content strategy deliverables

First off, don’t panic: The content strategist is here to help. (No content strategist? Considerdoing it yourself.) Jeffrey MacIntyre’s Content-tious Strategy is a practical overview of the various flavors of content strategist, and their respective deliverables; taken together, these documents are a gold mine for making smart CMS design decisions. Here are some highlights:

• An editorial strategy (“product development for content”) might include an editorial calendar, editorial workflows, and a style guide.
• A content analysis might include a content inventory, a gap analysis, a taxonomy, and a migration plan.
• Copywriting and IA-related gems include content templates (also called page tables), copy-decks, and annotated wireframes.

THE THREE TYPES OF METADATA

We haven’t mentioned metadata—commonly defined as “data about data”—because the term itself is confusing. In fact, Deane Barker argues that the distinction between data and metadata isn’t helpful. But let’s consider Rachel Lovinger’s definition (PDF link), which defines three types:

• descriptive metadata is taxonomy: classification systems for content;
• administrative metadata specifies behind-the-scenes status of content, normally managed by the CMS itself; and,
• structural metadata defines the content model.

Instead of using the same word for three distinct concepts, we’ll talk about taxonomy, administrative data, and the content model.

Content modeling: types, elements, relationships, oh my!

Based on the strategy, we’ll design a model to describe the website’s content: Types, elements, and relationships. You can think of a content model as a semantic structure for content, or a database schema; it’s part of the information architecture. (Don’t confuse it with the site map, which specifies top-down navigation.) Content modeling isn’t a straightforward, mechanical process; it requires human judgement and experience, and there’s no single correct solution.

BALANCING SEMANTICS AND GRANULARITY

Content modeling is about striking a balance between semantics and granularity. We can encapsulate it as the answer to two questions:

1. What does this content represent? (Semantics)
2. How much detail should we go into? (Granularity)

To demonstrate, suppose we’re designing a content model for a company that runs conferences. First, we’ll consider which content types are needed. (These aren’t media typeslike video or text: Each content type represents a separate entity in our model.) It’s easy to brainstorm possible content types: Events, presentations, speakers, attendees. But how much detail do we need? Should we model multiple conference “tracks” and schedules, or are presentations and speakers enough?

At the same time we’ll consider which content types are related to each other, and how. For example, if we model tracks, each presentation is related to a track; without tracks, each presentation is related directly to an event. We’ll also consider whether relationships are one-to-many or many-to-many (the technical term is “cardinality”). Next we’ll consider elements: What makes up each content type? In our example, do we need a “URL” element for each speaker, or is the “Biography” element sufficient? Finally we’ll decide which content types need classification, and which taxonomy to use.

DERIVE THE MODEL FROM THE STRATEGY

If this sounds a bit abstract, don’t worry. We won’t sit here all day debating the nature of the world; we have a content strategy to implement. Based on the deliverables outlined above, we might design this content model:

Each box represents a content type, and lists a few possible elements; the lines represent relationships between pieces of content. We’re modeling an event as a number of tracks, each featuring a number of presentations, each presented by one speaker. This semantic richness gives us the flexibility to present content in powerful ways. For example:

• From the speaker biography we could link to their presentations at past events.
• For each presentation we could automatically show what’s on before and after, and what’s happening simultaneously in other tracks.
• Search could return intelligent results, e.g., a speaker and the events they’re due to present at.
• There’s scope for a personalized conference schedule that attendees can use to plan their day.

DON’T DESIGN THE PERFECT MODEL

Consider a conference you’ve attended; does this content model make sense for that conference? It probably doesn’t: What about events with a single track, or panels with several speakers? Or more significantly, what if the editorial strategy is based on publishing high quality videos of presentations? We’re not striving for the ultimate model; we need a pragmatic design that accommodates the real-world constraints of the content strategy. We won’t get it right first time: content models evolve, so we’ll allow for iteration and change over time.

Task analysis: what’s an editor to do?

But a model isn’t enough; an editor needs to create, edit, publish, and care for the content that lives in it. In real life, the editor will be short of time. Using task analysis we can make our content model more realistic by considering feasibility. It’s a great way to identify assumptions about the CMS interface upfront, and it lets us inform the project team about the true cost of content and features—taking into account ongoing editorial and technical time.

WEB EDITORS ARE USERS TOO

For web professionals, task analysis isn’t new. We’re always thinking from the user’s point of view, making tasks as straightforward as possible. But how often do we apply the same type of thinking to web editors? Here’s a four step guide:

1. Brainstorm key tasks (based on the content model, editorial calendar, and content inventory).
2. Sketch workflow diagrams for each task.
3. Sketch wireframes of key interfaces.
4. Estimate the editorial time required to complete each task.

AN EXAMPLE TASK ANALYSIS

Continuing with our conference example, how does our content model stand up to task analysis? It’s straightforward to list an editor’s key tasks: Publishing a news item, adding a speaker, adding a new presentation, etc. Here’s what the CMS task flow might look like for adding a new presentation, given our content model:

The diagram shows the five processes and two decision points involved in this task. A typical implementation might include five separate screens; we’ll estimate that an editor will take 10-20 minutes to complete it. We’ll also sketch simple wireframes for each process. (Would an Ajax-enabled design work better? We’ll need detailed sketches showing how the auto-complete or show/hide magic works.)

If we apply task analysis to the whole system, we can derive sensible estimates of the editorial time required to complete each task. We can then prioritize our scope based on the actual time available. It might turn out that parts of the content model are over-ambitious, while others need to be extended. This process helps us find a realistic balance between modeling and task flow based on strategic priorities rather than haphazard assumptions.

There’s another benefit: Identifying assumptions within the content model about the publishing process. For example, our task flow requires an editor to select a track before adding a presentation. Is that a valid assumption? What if we need to publish presentations before the tracks are finalized? And how will the front-end website present tracks anyway? Finding these assumptions before implementation saves time, money, and grief.

Decision time: CMS selection and customization plan

At this point, we have a revised content model, and a task analysis that specifies how editors will interact with the CMS. We also know which tasks are most important, which will help us to prioritize backend interface design work. This puts us in a strong position to shortlist and select CMS tools, and to scope customization.

There’s no silver bullet for CMS selection. The key is to specify the problem as clearly as possible, and then to insist on realistic time estimates for customization and implementation. We’re ready to ask the following questions:

1. Can this tool handle our content model? Natively, or will it need customization?
2. How much customization will be required to implement this task flow? How long will that take, given our technical resources?

If you aren’t a technical expert, you’ll need to consult your technical team, vendors, or online communities. Although there are other important factors to consider when selecting a CMS (e.g., platforms, licensing, hosting), don’t let anyone use them as an excuse to avoid answering these basic questions. If the tool can do what we’ve outlined, and there’s enough time and money to customize it, great. If not, we either need to consider a different tool, or scale down our plans so they’re feasible. A CMS project needs technical resources after launch day too, so be sure to get estimates for ongoing design changes and maintenance.

The output of this process is a project plan. We know which tools we’re using, and we’ve scoped the work required to get them to fit our needs. Now, buy yourself a drink.

That final arrow: back to the drawing board

We’re not quite finished, though: We can’t just pour in some content strategy, customize the CMS, and leave. CMS design is part of an ongoing, iterative process: Our selection and customization plan provides information about feasibility which will influence the content strategy itself. In practice, we’ll dream up ambitious plans which require unrealistic amounts of editorial and technical time to implement. So we’ll use the workflow diagram’s final arrow, go back to the drawing board, and scale down our plans to make them more realistic.

REGAIN CONTROL OF YOUR CMS

Our websites have been held back for too long by lackluster publishing tools. In this article we’ve explored ways to apply strategic thinking to CMS selection and customization, through design. It’s time to regain control of our content management systems by harnessing the power of content strategy. 

About the Author

Jonathan Kahn is a web developer, content strategy advocate, and basmati rice aficionado. In 2008 he founded Together London, a collaborative web design agency. Jonathan listens to soul and jazz, and writes at lucid plot.