This Is a Useful Approximation

Content models that are wrong but useful

Hello, fellow traveler. Every week (or so!), this dispatch helps you make better creative and strategic decisions. Sometimes that means essays. Sometimes that means resources. Last time it was a brief meditation on the anxiety of the internet and how we organize our minds. This time it’s a brief essay on models and a few content strategy slides to download. Anyway, hello, it’s good to see you. -Steve

Gif by Martin Tognola


In the field of statistics—a field about which I know absolutely bupkis, except the single fact I'm about to relate—there's a famous aphorism that goes like this: "All models are wrong, but some are useful".

The statistician who coined that aphorism was George Box, a gent who founded the stats department at University of Wisconsin–Madison and, before that, spent World War II performing experiments exposing small animals to poison gas. (I didn't know that either, but Wikipedia did.)

The point of Box's aphorism is that models are simple representations of complex realities. They are by their very intent only approximations. What we must pay attention to, Box said, is exactly how approximate the models are.

As an example, consider the ideal gas law (Box was really into gas). The ideal gas law is an equation that describes of state of a hypothetical gas.

"The law PV = RT relating pressure P, volume V and temperature T of an 'ideal' gas via a constant R is not exactly true for any real gas," Box wrote, "But it frequently provides a useful approximation and furthermore its structure is informative since it springs from a physical view of the behavior of gas molecules."

In other words: gasses don’t act exactly like the model suggests, but the model is close enough to be useful.

We’re surrounded by models

It so happens that we're surrounded by models.

Take maps. Maps are models. They are symbols emphasizing relationships between elements in space, they are not the space themselves. Hence the expression "the map is not the territory".

Consider the New York Subway Map. The New York Subway Map emphasizes navigability at the expense of geographic exactitude. The simple subway lines are not the chaotic streets.

Or consider The Knowledge of London. The Knowledge of London is a test that requires every taxi driver applicant to memorize the city's 25,000 streets and landmarks. But that feat of memorization, impressive as it is, represents only an approximation of the city—not the city itself. (Regardless, the 1979 comedy about the test, starring the future director of Clue, is delightful.)

Even language is a model. "The first thing you learn in general semantics is 'the word is not the thing'," begins an old William Safire column. "That is, the referent is not the source, or, to put it understandably, cow is merely a word, a sign, a name, and is not the flesh-and-blood thing that has an udder and goes 'moo.'"

Or take the idea of god. Jesus is a model. Allah is a model. The Flying Spaghetti Monster is a model. The models, as representations of reality, are likely wrong, but they're certainly useful in that they give people a manual for living (or at least, in the case of Pastafarianism, a tool for opposing the teaching of intelligent design in Kansas state schools).

Since we've now arrived at the realm of the philosophically abstract, let’s just go ahead and consider models within the most abstract practice of all: marketing.

Marketing is an entirely conceptual space filled with incorrect but useful models. Arguably the most popular model you’ll find in this breezy land of bloviation is the purchase funnel—i.e., the model of how customers buy products. The purchase funnel is useful in that it helps a business develop stage gates and reinforcement mechanisms as consumers consider, evaluate, and make purchases. It's incorrect in that I buy things on a whim when I'm sitting on my couch two drinks in.

You're wrong, but the way that you're wrong is useful

In a business marketing context, you have to be careful and deliberate when you deploy models. What's useful in the abstract can be much less useful, or even distracting, when it’s applied to a brand that’s already in motion. I'll give you an example from a gig.

Recently, an agency asked me to develop a content strategy for a brand. The brand was developing an app. The app would have editorial content. What do we need to be thinking about, they asked, and how does content work? I told them that creating content was a complex but manageable task that required creating production and governance processes across departments. To illustrate, I showed them the content strategy quad:

The quad is an excellent model. It reveals the high-level components of content strategy and contextualizes those components within the practices of content design and systems design.

Editorial and experience live within content design—that is, how you develop content to meet the expectations of your audience. Structure and process live within systems design—that is, how you create repeatable systems to create and manage that content. As models go, the quad is comprehensive and elegant. I recommend it and use it often.

So during a meeting I pointed to the quad and said these are all the things you must consider. They took a look and said this is great! And then they took a longer look and said: but what do we do with it?

No model survives contact with the enemy

In an ideal world, you can be deliberate about staging projects. You can structure step-wise presentations to help your audience build conceptual knowledge. This becomes tougher when a business is already in motion. Time pressure to execute creates the need to move quickly.

In the case of my gig, the brand needed more than to understand “how content strategy works”. Rather, they needed several things at once, including:

  • What is content strategy?

  • What is our content strategy?

  • How will content strategy affect our project?

  • How do departments work together?

Those needs don’t align neatly to abstract conceptual models. For my client's purposes, the content quad wasn't wrong but it was too approximate to be useful in the way they needed it to be useful. What they needed, instead, was content strategy in the context of policy. That is, they needed a different type of approximation.

Simple models of complex context

The first thing I needed to do was to illustrate how strategy trickles down into several disciplines. Or in other words: why are we doing it, what are we doing, and how (this won’t show up very well in mobile, but there’s a download link below):

That’s️a good high-level slide. It helps its audience understand that strategy has downstream implications across disciplines within its business. This is the kind of context that many content models lack. That’s not a criticism of the models, but an acknowledgement that businesses don’t operate in the abstract.

This strategy though, once understood, will need to be made more operationally explicit. For that, here’s a version showing broad responsibilities across departments:

Each of these models are generic. Some projects have more departments. Some projects have fewer. Some projects, bless them, don’t have a legal department. Feel free to remix these models yourself. You can download both as Google Slides right here.

They’re not rocket science. But I hope they help you organize your own projects.

Just remember: these are models.

They’re definitely wrong.

But I hope they’re useful.


Last letter’s most clicked links:

  1. This is the best font guide I’ve ever seen

  2. A Surreal New Bookstore Has Just Opened in China

  3. Concepts is an app that offers flexible, infinite sketching


How can I help you?

This 100% organic, free-range, desktop-to-inbox newsletter is devoted to helping you make better creative decisions in marketing and beyond. Delivery at 6pm ET every Sunday, sharp as cheddar. Your host is Steve Bryant, who is for hire. If you’d like to chat just grab a 30-minute meeting. I’d love to meet you.


If you found this dispatch useful, please forward it to your friends and lovers. Thanks for hanging out. Be seeing you.