Experimental Content Strategy: The Snowflake Pattern
Exploring strategies for digital dissemination of media
Content strategy isn’t a secret anymore. Where once the notion of a “content strategist” was novel, the term has gained popular ground. It’s been adopted into our digital nomenclature alongside growth hacking and 10x content as buzzwords turned cliché turned institution. We practically take it for granted now.
Defining content can be as nebulous as ever. It can be a blog, or a video, or a tweet, an Instagram post or story, web stories, a movie (if you’re a streaming service), an article, a sponsored content post, or anything else that fills and fuels the demand for more content, spread often across many channels. Everyone these days has a pipeline to fill.
Of course, narrowing down content needs and audience needs is tougher. It involves heavy research, not only of relevant SERP results, but also of similar content that might branch or connect thematically without getting as much notice. Great content pulls from many sources to fill more needs. Copying a post and adding your own spin doesn’t cut it anymore. One blog on a topic, decently written, doesn’t get the traction it did fifteen years ago. Creativity has extended to the structure as well as the content.
It often helps to visualize content in different ways to make it work for different methods or goals. We’ve seen broadcast models for content, mimicking television and theatrical film distribution. Those are certainly tried and true formulas. A creator makes a message, disseminates it down to followers, who then share it with their applicable friends or connections. Makes for good watercooler talk, once consumed.
Some have even ventured into the very useful category of information architecture, or structuring content like a website for maximum findability. Internal linking has become essential for content structure; you want your readers to naturally find and follow the content that pertains to them, and hopefully fits their stage of the content funnel and buying journey. You’ve taken the trouble to tailor keywords to the segmentation funnel, now you need to make sure it hits that audience.
But what if we looked to other sources?
In The Stack, Benjamin Bratton compares internet fragmentation and digital production to the treaties of Westphalia, which established sovereignty and distinct boundaries for nations of Europe. Borders are power, land is control. He then applies this to other layers, such as city, state, and planet, to form a picture of ubiquitous computing at relevant scales. Bratton’s interest, and our most helpful takeaway, is the application of geography and physical space to cyberspace. Maps and territory can exist in bits and bytes.
Being from a geography background myself (urban design, not just memorizing capitals and rivers) the application of this discipline to our content structure posed many interesting ideas. We have known continents, such as Facebook, Twitter, LinkedIn, etc., where massive groups of people conglomerate, share, interact, and form economies and ecosystems. We know these lands well; we typically live there ourselves. As interoperability hasn’t occurred between them yet, these continents have firm, Westphalian borders of digital sovereignty.
But there are dark continents, or lands we don’t know about. Dark social media, or private messages and emails, or communications shared socially that we aren’t privy to, are the undiscovered and unexplored continent. We know it’s there, we’ve seen the blank space on the map. But we can’t see the land itself. We can’t study their conversations, we can’t reach them in their private digital operations. For every public post there is a shadow, and in that shadow lurks the trove of information. That trove has content galore.
We’ll explore the notion of digital sovereignty and land as internet in another article. It deserves its own. But what if we went further? What if, instead of just looking at maps and borders, we began to look at shapes and objects in the known world? What if we mimicked our design and our strategy on structures observed in nature?
There is a long and healthy tradition of examining natural fixtures for clues as to the self-organizing powers of the environment. Fractals are structures with similar patterns that recur progressively at smaller scales. They describe chaotic or random phenomena. You’ve likely encountered them before, such as the Fibonacci spiral. Fractals and their application to other natural phenomena are a particular focus of the Santa Fe Institute, which is a breeding ground for some of the best research into complexity science in the world.
A snowflake is a beautiful example of a fractal. Made of crystals that display complexity that increases with magnification, a snowflake encapsulates what we love and wonder about in nature: how did it come to emerge and look this way, and why has this underlying structure stood out over time? What about it appeals to us? What about it is universal in its design?
And what if, for our experimental content strategy this week, we mimicked it?
This snowflake is composed of a central hub, surrounded by branching nodes connected by edges. It is designed to replicate a basic snowflake pattern, although, as we all know, there are innumerable different templates one could draw from. Increasing complexity followed by granularity increase the fractal nature of the design. Replicated and repeated content structures mimicking the main design increase the complexity and the resemblance to an artifact from nature.
The main hub is the pillar, or anchor, content. It can be an article, blog, or video. Whatever it is, it needs to contain valuable keywords and be longer form, or at least in depth enough to cover a topic or a series of related topics very well. For example, one could have a 3,000-8,000 word blog or article about a given topic as the hub. A longer video including external links to sources in the description for more information would branch it out as well.
From there, nodes are thematically related bits of content linked to the hub via detailed metadata strategy, internal architecture, and hyperlinks and anchor text. The nodes and their content do not have to be the same medium, but they should serve the same class of customers and/or serve the same functional content needs. This helps to distinguish them by purpose. If one is using the snowflake content strategy for an ecommerce site, then the nodes should equally be in the same part of the buyer’s funnel. Nodes should have roughly equal weight.
A node can be a YouTube video that talks more about a given subject in one of the paragraphs of the hub article, for example. Another node might be a shorter blog or an article on a different social platform or publishing medium that goes into depth about a keyword cluster featured in the hub article.
A good metadata strategy not only encompasses keywords and phrases, but also keyword clusters and their semantic connection to the rest of the material. A controlled vocabulary or a specially created and specific thesaurus is always a good place to begin. Creating and using synonym rings for keywords and phrases for use across hubs and nodes certainly can’t hurt; neither can being aware of different categorization languages and standards, such as RDF or XML.
From there, the snowflake pattern is replicated further down. One can branch the overall structure from here or increase granularity depending on need. For example, the next hub might be a shorter article or different video loosely tied to the first, which then is connected to tweets or Facebook posts as nodes. Again, nodes distributed in this fractal way should serve the same purpose for ease of translation. Eigenvector centrality, or the weight of nodes and their relation to others, should start out roughly equal to one another in this system.
All nodes and the hub should be connected not only by direct hyperlinks and anchor text, but also by inferred links. This counts across the entire snowflake; anytime there is relevant linking for internal navigation, a hyperlink is valuable. However, part of this structure mimics nature in the fractal principle. If, for example, the content strategy includes a robust series of content pieces about a given topic with related subtopics, then merely discussing those topics in smaller node posts will likely automatically link to the main hub via inferred links. The chances of inferred links being created and growing increase as the amount of content increases. It scales down continuously while remaining thematically connected and branching.
Snowflakes can take on any level of desired granularity or fractal repeating. If the topic is related to the core functions of the business, then very likely the content will always be linked and snowflakes will grow into one another and cross over. If the pattern is held to, then soon thousands of hubs, edges, and nodes will intersect and branch in a giant network of content, repeated as fractal patterns if one were to precisely map the content across the platforms and their distribution channels.
Of course, this is a lot of content. Whereas in nature snowflakes are a self-organizing fractal property, content is made by human hands. Each piece has to be crafted and generated by a person, written, edited, filmed, etc. At least, so far. In the future, it is highly likely deep learning and AI will lead to the ability to automatically generate content that precisely mimics the hubs and nodes scale. It will then be able to churn out massive amounts of content and fractally increase the complexity and its branching structure.
If one were to design a machine learning system that generated equal-weight node content pieces, such as two blogs and an article published on separate digital spaces, then content would indeed become like a snowflake, so indelible to the internet and its structure that it is a self-organizing property created by the environment that came before. Content pieces are the atoms and the molecules, connected by inferred links and text as bonds that increase complexity beyond a human scale.
Our entire digital ecosystem could, theoretically, be automatically generated and looked over by a few content stewards who ensure the machine self-replicates, copies, and creates new material. This fulfills the old joke from Warren Bennis about the factory of the future: “The factory of the future will have only two employees, a man and a dog. The man will be there to feed the dog. The dog will be there to keep the man from touching the equipment.”
That being said, it’s only a possible future. It might be that human hands will always fashion the things humans desire to read and engage with. It might be that a mathematically branching snowflake created by an AI is too perfect, like a fabricated city. It won’t have the brushstrokes we’ve come to identify as human. It would lack apparent age. It would lack the messy quality that reads like a person was behind it. It might link to the wrong place, or send them off-message without even realizing it.
Perhaps there would be a mishmash, a combination of human and creation. Maybe the mathematically perfect and weighted hub and node system, created by deep learning, will only be the frame onto which humans then craft branching pieces themselves, of equal or not equal weight. The perfection of the system gives way to entropy, the structure breaking down the way the internet naturally does. For instance, remove one piece, and you have link rot lurking in another content piece; change anchor text or edit a piece substantially, and you might lose the inferred links and their resultant node weight.
As a viable content strategy, the snowflake should be pursued as a starting place. It gives a shape and a pattern to the distribution of pieces, and it helps show how metadata link the items together. It shows the way people might digest the information across the hubs and their nodes, and how customers find the products they’re looking for. It should never be forgotten that the prime purpose of the content strategy is to appeal to the customer. This is designed for them. The snowflake is a possible template for how one might produce work naturalistically to appeal to our sense of the world around us. But every piece, every node, should be designed with usability and purpose first and foremost.