You didn't discover a new way to write

How many ‘GEO’ specialists have you heard say recently that you need to write a summary before you dive into anything ‘too deep’ in your article? Or that you should break information up into chunks? Use bulleted lists? Keep it to one-concept-one-paragraph? Use clear, descriptive, headings without the brand ‘fluff’?

Quite a few, I’d bet.

And worse, they probably present it as a new concept.

It’s not. The sub-editor at the news desk on the other side of the building could do that in their sleep. And probably does, sometimes.

The Paper Trail

Telegraphs were a formative experience for newspapers, or so folk like to say. It actually created a style, dictated by necessity. Short, punchy. Stripped of any unnecessary detail — usually the opinion. Why? Because like early text messages, telegraphs cost.

But actually, the inverted pyramid of ‘key facts first in case the wire was cut’ is mostly a myth. The style actually came through in the 1880s, for a more pedestrian reason: the desire for clarity and comprehension. Sound familiar?

Then this evolved further with the summary lead and the nut graf. (Yes, I know, we all have a teenage boy’s brain. Have a giggle).

And the nuance here:

• Summary lead: the w’s — who, what, why.
• Nut graf: so what? The hook.

GEO flattened the nuance. Journalists argued, and continue to argue about the role of each. Take a look at the Pulitzer-winning “Snow Fall” article. And a number style guides warn against oversimplification of the inverted pyramid and its associated methods.

Readability

You may have heard the name Flesch before, in fact, I hope you have. Tools like Screaming Frog use Flesch’s methodology from the Journal of Applied Psychology in 1948 to generate a readability score. It’s also baked into Microsoft Word. Yoast too, or at least it was, once, in the guts of it.

The readability standard created through this methodology advocates for short sentences, short paragraphs, and plain style. Or take the same advice from another, where Fowler said, in 1906, “direct, simple, brief, vigorous and lucid.”

The Internet of It

In 1997, when we were all still asking Jeeves, Jakob Nielsen found, essentially, readers don’t read on the web. They scan. Only 16% read word-by-word. He advised, back then: meaningful sub-headings, bulleted lists, one idea per paragraph and using the journalistic inverted pyramid to start with the conclusion.

And it was measurable then, which kills “but now we can prove it”: his five-version study scored the concise site 58% higher in usability, scannable 47% higher, objective language 27% higher, all three combined 124% higher. Obviously this was not about AI or LLM citations. And there was a 2020 update. Same findings.

Why? Because that’s just the way humans read. Regardless of the medium.

So here we are, in the age of GEO, with folks saying journalistic principles more than 150 years old are new. Yes, the translation of that to the technical interfaces with machines is, but the principles behind it most definitely are not.

But folks are trying. Someone decided to call the inverted pyramid the ‘ski ramp.’ Nielsen acknowledges the concepts are old, and folks are pulling in journalistic concepts, like (not) burying the lede. Not enough of us, though.

We don’t need a new playbook. We need to read the one from 1997. Or 1948. Or 1865.

A Technical Qualifier

There are two things happening here.

One, the content machines have been trained on: a lot of it was written by journalists trying to keep the attention of readers, using the inverted pyramid, the summary lead, the nut graf. There is precedent already for this kind of scannable content in the LLM training set. So it often gets generated back.

Two, an LLM’s preference for the beginning of a piece of content is the ‘lost in the middle’ effect, a side result of the transformer’s positional encoding.

So regardless of whether the vehicle is new — which, yes, it is, partially — the way we solve for it is still the same way we solved for human inattention for a long time. And this matters, because there should be no panic here. This isn’t a new discipline, out of thin air. It’s an old one, that SEO hasn’t been fully utilising.

Don’t believe what you’re being sold.

And honestly, question the folks who are selling it to you. If they don’t acknowledge or understand the journalistic history of the shape of the organisational concepts they’re discussing, I would question their understanding of the marketing discipline as a whole.

We’re at a point where you can’t be a good, low-risk, strategic SEO (or GEO, or AEO, or whatever) without…just being a marketer.

So here’s the kicker.

We didn’t lose the craft. We never hired it.

Journalistic principles aren’t a lost art. They’re used every day. It’s your stressed coworker Suzy, three rows over who is trying to manage the content for a 100K page website with 4 writers and a lazy legal team.

If you told her “make sure to use one idea per paragraph” she’d give you a dead stare and pour her coffee down your shirt. Cause that’s week one shit. Autopilot. You don’t get a byline if you don’t know that type deal.

As SEOs, we decided to reinvent all these concepts, badly. From blog posts written sometime in 2012 during the last programmatic content explosion, rather than talking to the people who did this for a living.

So why didn’t we talk to them?

Not the simple answer. It wasn’t purely that we were “writing for search engines instead of people.” The advice has been to write for humans for years, since Panda. Earlier. It started with exact match keywords and the density of it was rewarded.

But there are larger, institutional considerations. Editing doesn’t scale. It’s not cheap. For a long time, if a few paragraphs were messy on the website, your rankings didn’t care. But at the other end, aiming to churn out 50 articles a month is terrifyingly messy, too. US newsroom staff fell 35%, 2001–2014, copy desks hit hardest; the NYT killed its standalone copy desk in 2017. The same cheap-scale logic gutted the newsroom and built the content mill.

It’s also hard to measure. How do you quantify something being “well-edited”? Or trustworthy? So rather than focus on the how or the why of the writing, we focused on the sessions it drew or the keywords it ranked for. Easier measures, but not a proxy for the deeper value of writers and editors and journalistic style.

Content became a commodity in an industry long dominated by STEM rather than STEAM. Which meant we skipped a lot. Things writers will do without being told twice:

• The editing pipeline: structural edit, line edit, proof, a second pair of eyes. Organised around Accuracy, Brevity, Clarity (Journalism University); what separates professional journalism from casual content creation.
• Fact-checking and the correction: The SPJ code: verify before you publish, “neither speed nor format excuses inaccuracy,” and correct mistakes “promptly and prominently.” Name the last time a brand blog ran a correction. We don’t fix. We re-optimise and move on.

And editing has measurable reader value: readers rate edited stories higher on quality and professionalism, even lay readers with no journalism training. The thing we cut is the thing readers can feel.

What I’m not saying here: hire a freelance journalist to write your content.

What I am saying: we need to bring (back) in this thinking, this rigour, this structure.

We never had the craft, and rather than pretend it didn’t exist for the 150+ years it’s been in use, we should leverage it, learn from it, use it ourselves.

So what does that mean?

1. Add the editing layer. Structural and line passes, done by a buddy, someone not the author.
2. Fact check. Verify it all. Every number, name, and claim. Against a primary source. Then correct if you’re wrong. Clearly.
3. Reallocate budget from volume to expertise. Invest in talent, not churn. Hire an editor. Bring in standards, or bring them back.

Everyone GEO guru says the machine rewards clarity, structure and trust. Sound familiar? The bet here isn’t “do journalism to win the algo.” It’s far simpler: write the way we’ve always written for people with short attention spans.

The upgrade isn’t a better prompt. It’s a copy editor.