Illustration by Sir John Tenniel

When I was young, I started out as a reader. I read stories, and read them (usually) from start to finish.

When I began to write, I learned that different types of writing had integrated patterns - ideas could be constructed through argument, outline, connections and conclusions; or by listing the facts, beginning with the most significant detail.

But not every pattern worked for every subject, everywhere, in the hands of every writer.

How we write now

For the most part, I create content strategy for large-scale commercial websites. Maybe you do, too.

The writing we’re involved in is workhorse writing. It exists to get people where they want to go, or to tease them along til they’re at the point of doing something they didn’t even know they wanted to do. It runs the gamut from coy flirtation to outright nagging.

But most of us still want our words to be, in some empirical sense, good. We operate on the principle that there is an attainable ideal for the words we create. Content that voices a brand perfectly - and that reaches people.

But what makes good writing good?

Strategies for words

When we create a content strategy, we implicitly say that by planning for content, we can create patterns for our writing in advance.

But content strategy is not all-seeing - it’s a framework. When we create content, we do so knowing that as soon as those words are published, they must be measured and changed.

In order to write effectively, we have to consider every word. But online, our word choices ought to be as much about structure as they are about style. Change a key word, and you change the whole experience.

Writers are latent control freaks

When putting together a strategy, I look to analytics for widespread patterns and repeating problems, backing it up with human analysis - audits, comparative reviews, common sense.

But when a site launches and passes back into the hands of a marketing team, content often starts to be viewed in a new way - not as part of a cohesive structure but as a small piece of a wider, but often shallower campaign.

Don’t get me wrong, this isn’t the wrong way to view content, but it’s incomplete. It takes a very narrow view of the roles content needs to perform, and pushes through changes without thinking about structure - risking inconsistency and incoherence.

Data is not self-aware

Optimisation is a familiar concept for a writer - the idea that there is a perfect marriage of words and images out there, an optimal ideal we can aim for.

But just as a writer can never be sure they’ve found the perfect word, when we optimise content, we’re never selecting variables from a closed set. There are countless possibilities.

Words are responsive - not just to design, but to one another. When you edit a piece of writing, you remove elements that aren’t working - not because they’re wrong, but because they don’t support the whole structure.

In the same way, optimisation requires constant, careful evolution. It requires us to be responsive, rather than reactive.

It also requires us to think about what’s happening behind the data - such as:

  • Who are these people, anyway?
  • What kind of content have they responded well to in the past?

Pattern recognition

Pattern in data is useless unless we take the time to identify, analyse, and do something with it.

Which brings us to question 2, and my point.

Too often, I see optimisation exercises that see each page and piece of content as a completely new unit of information - instead of building on knowledge from the past, we throw a bunch of variations at the wall and see what sticks.

Then, next time, we do the same thing - rather than looking for patterns in the type of phrasing and word choice that succeeds with our audience.

If we:

  • Track content’s journey through optimisation - not just about which page won, but what the nature of each winning piece of content was
  • And if, before that, we base our decisions for optimisation on a clear, coherent content strategy
  • And ensure content changes are made by writers and editors, not by marketing committees

… we could start finding that the overlap of content and data can do a lot more than we’ve been asking of it - that it answers, and refines, our content strategy.