Issue 43

Your content is scaling. Now what?

I live and breathe content architecture, and if you don’t believe me, listen to this: last night I dreamt about scaling content.

I’ve presented myself as someone who can help scale content. Recently, I’ve seen a number of job postings that task software engineers with creating scalable content models. Maybe more so in the past, I’ve seen content teams scaling up their staffing. And we know that the amount of content itself is scaling up.

This has me thinking.

Firefighters learn to “read the smoke” as a way to know what the fire is doing and will do if not dealt with quickly.

They learn the mnemonic of “volume, velocity, density, and color.” For example, increasing smoke volume and speed (velocity) of the smoke (from lazy to roiling) are signs of an intensifying fire. Likewise for wisps of smoke changing to dense columns of smoke. Smoke changing from gray to brownish means it has moved from its starting stage to involving the structure (e.g. studs and rafters) itself—or from underbrush to trees in a wildland setting.

Reading the smoke: signals that your content is scaling

If your organization has content that is scaling, you need to recognize that and read the signals, just as a firefighter would with smoke. The scaling content mnemonic might be “volume, velocity, semantics, complexity, and staffing.”

Volume

In recent years, humans have churned out more and more content. I’ve seen this happen at numerous organizations too. And now we have synthetic content being spat out by large language models (LLMs) at rates that are mind boggling.

The more content your organization publishes, the greater the challenge to manage and the more risk you’re exposing yourself to.

Velocity

I come from the world of weekly newspapers. We published one issue that came out every Wednesday, and that was a pleasant cadence in rural Virginia. Other newspapers in our chain came out daily—and still do, although now they likely publish news to their website as soon as they can. The velocity of journalistic content has increased to near-instantaneous levels.

When I started in the software industry, we shipped software releases on CD-ROMs. I was a technical writer, so our documentation had to be part of those releases. I don’t remember the exact cadence of those releases, but it was only a couple times a year—quarterly at most. Then, I think we started shipping downloadable updates quarterly or every six weeks or so.

Now, much of the software we use is cloud-based software that is shipped with “continuous delivery” methodologies, so user assistance content could be shipping daily.

Regardless of your industry, my hunch is that the cadence of your organization’s content is increasing. Your risk is not having governance or infrastructure to keep up with the new velocity.

Semantics

Whether you’ve realized that your front-end design-based content model doesn’t work or you’ve learned that the “semantic layer” is one of the keys to win in the age of AI, your content is taking on more systematized meaning—semantics.

You might think of this as a maturity model. You start by moving from modeling presentation structures to modeling meaningful content. Then you start defining meaning itself through controlled vocabularies, taxonomies, and ontologies. Lastly, you lean on this work as you connect it to systems—AI, search, chatbots, etc.—that need to understand that meaning.

Your risks here are locking content to specific presentations and experiences, assuming that data modeling and content modeling are exactly the same thing, making your taxonomy too narrow, and failing to show your organization how the semantic layer unlocks new capabilities.

Complexity

There’s a lot of ways content gets more complex:

• Adding in a semantic layer.
• Adding placeholders/variables for multiple different outputs (like “Click [X] to advance” where the cloud-based users get “Click OK to advance” and the self-hosted users get “Click Yes to advance.”)
• Adding variants to enable personalization.
• Changes to the content management systems (including CMS, DAM, and other systems).
• Implementing governance.
• Having more people working on it (see below).

Some of this complexity involves content architecture and systems, and some of it digs more into content operations. It’s easy to say “AI can do that,” but if you say that, you need to then ask questions about how to systematize and make the processes repeatable.

If you say, “AI can do the personalization variants,” yes … AND … what system will store those variants? How will it store them? As true variants or as new content entries? How will the system know which ones to display at which times? How can multiple team members work with the AI and the variants?

Here you risk underestimating the complexity of changes, or assuming that a magic wand can solve all the complexity.

Staffing

When you’re a solopreneur doing all the content things, you accept a lot of imperfect processes and tools and do some pretty creative things. But once you bring in another person to help you with the content, you start to see fragility if not outright problems.

Same is true as startups grow and start bringing more specialists to help with different parts of the business. And it’s still true as mid-sized companies grow into enterprises.

Organizational growth begets content operations breakdowns and difficulties.

If you fail to recognize how more contributors and collaborators changes the dynamics around content, you risk chaos and content anarchy.

Interpreting and responding to the signals you see about scaling content

When firefighters read the smoke’s volume, velocity, density, and color, they don’t always employ the same tactics on every fire.

Sometimes they will (don’t be surprised here) put water on the fire to cool it. Sometimes they might remove fuel so the fire burns itself out. Sometimes they will open doors and windows to change how the fire behaves (and access the fire), but other times they know that enlarging an opening will lead to the catastrophic reaction known as backdraft.

Likewise, when the content professional sees the signals of volume, velocity, semantics, complexity, and staffing, they know that multiple factors are at play, and the response needed may vary with some mixture of content strategy, content architecture, systems, and content operations.

Volume

An increasing volume of content could mean many things. Maybe there’s new products, acquisitions, or the content team is catching up with a backlog of work. You might want more content in those cases.

Possible responses to increasing volume include, reducing the amount of content, improving the CMS and related systems, increasing content capacity, and improving the prioritization process for content.

Velocity

Increasing content velocity isn’t necessarily a bad thing. You may be delivering more user value more quickly than ever before.

You might choose to respond by controlling the flow of content, improving governance, increasing automation, or focusing on workflows.

Semantics

More semantics attached to your content is a good thing. You’re on the right path.

When you recognize the semantics signal, your response should be to resist panic. Firefighters borrow from the military and live by the mantra “Slow is smooth; smooth is fast.”

You too should go slow to go fast. This is a good time to step back and have the hard discussions and do the diligent work to get your content models or taxonomies ready for the opportunities ahead.

Complexity

In a certain sense, increasing complexity is the cost of doing business. You can expect complexity when enabling new capabilities, but content-related complexity can also stem from strategic gaps or flaws, from content architecture problems, or from operational gaps.

When you recognize that your content is getting complex, responses might include clarification, simplification, standardization, or accommodation.

Staffing

Like many of the other content signals, increasing staffing around content isn’t necessarily a bad thing. In fact, I’ve often made the case that there are many roles missing in the content professions (such as editors, content architects, personalization strategists, to name a few).

Respond to increased staffing by clarifying roles and responsibilities, changing organizational structures, or improving tooling.

Closing

As a firefighter, I watch a lot of helmet camera or bystander footage of fires, training myself to read the smoke. As a content professional, I watch organizations wrestle with scaling their content to the point that, apparently, I think about it in my sleep.

In both cases, the skill is in recognizing the signals and choosing the appropriate action.

Information Architecture For the Web and Beyond by Louis Rosenfeld

Top of mind

• Best thing I read in the last 2 weeks: Patrick Neeman wrote an article with a pull quote that really resonated for me: “Poor information architecture can now be measured in token costs. A lot of them.” Give the whole piece a read: Information Architecture Is the Foundation Artificial Intelligence Is Starving For.
• Something that made me smile in the last 2 weeks: I’ve written before about the local professional soccer team, Austin FC. Last week, one of our strikers scored a hat trick (3 goals) in the first half of a tournament game. It was the second hat trick in the club’s history, the player who scored it struggled last year, and all three goals were different and thrilling.
• A pattern I noticed recently: There’s nothing deep about this pattern recognition. It’s probably not really even a pattern. About this time for the last three years, I’ve made a road trip that involves coming into Amarillo, Texas from the northwest. Each time I’ve done this in the last three years, I have driven into rain on the same short stretch of road on the outskirts of the city.

Welcome to the 2 new subscribers who joined us since the last issue of Model Thinking.