Your SEO Formatting Triggers Inauthenticity Aversion

You open a blog post in a tab while you’re already half-distracted, and within two seconds your eye has done a survey. H2 under the title. Three H3s further down with bold lead-ins. A callout box. A “Key Takeaways” panel near the top. An FAQ section anchored at the bottom. Em dashes scattered through the paragraphs like a tic.

You haven’t read a sentence. You have already closed the tab.

You couldn’t tell yourself why. The post might have been useful. The writer might have been someone you would otherwise want to hear from. What ran in your head wasn’t a verdict. It was a recoil. Something about the shape of the page tripped a wire underneath conscious thought, and the page lost before its first paragraph loaded.

I have watched myself do this to writers I respect. Worse, I have watched it happen to writing I helped produce. The shape gave the work away before the work got a chance.

That recoil has a name. Researchers studying online deception call it inauthenticity aversion, and it lives close to the neurobiology of betrayal aversion. The brain registers a pattern of inauthentic signals (visual, structural, rhythmic) and hard-recoils, the same way it recoils from a smile that’s almost right but not quite. Once the signal lands, no amount of good writing rescues the page.

The shape of your content is now the first sentence of your content. And most of you are writing it in a font you didn’t choose.

How we ended up writing like the machines

For fifteen years, the SEO industry sold every content team in the world the same instruction sheet. Use H2s every 300 words. Lead bullets with bold phrases. Add a TL;DR at the top. Include an FAQ schema at the bottom. Place a numbered list above the fold. Keep paragraphs to two or three lines for scannability. Generative Engine Optimization tightened the screws further: now the page also has to be readable by a model summarizing it back into a chatbot answer.

The optimization worked. The pages ranked. The pages were also visually indistinguishable from each other, and from every page an LLM now generates by default.

This is the part most content marketers haven’t reckoned with. The instruction sheet you have been following since 2012 produced the exact visual fingerprint of a machine-written post. Not because writers were lazy. Because the rulebook *was* the rulebook, and the machines were trained on people following it.

When ChatGPT writes a blog post, it doesn’t invent the shape. It hands you back the shape your industry already standardized.

The tells live above the text

The thing I underestimated, until very recently, was how much of authorship reads at the level of layout. The audience isn’t absorbing your argument when they bounce. They are parsing your skeleton.

Turnitin, the academic-integrity company whose AI detector now scans student submissions at most US universities, spends every day looking at machine-written work and has been cataloging the structural tells. Vague attributions (”studies show,” “experts agree”) with no named source. Predictable three-beat cadences. Bullet lists that repeat the same syntactic shape four times in a row. Superficial_-ing_ analyses (”emphasizing,” “showcasing,” “fostering”) that announce thought without doing it. The tells aren’t in the topic. They’re in the architecture.

Any reader who has spent a year inside ChatGPT, Gemini, and Perplexity has been training on this architecture too. They are now fluent in machine-shape. They can see it in their peripheral vision. They do not need to know what triggered the recoil to feel it.

What a human-shaped page looks like

The fix is not aesthetic minimalism. It is specificity at the level of structure.

A human-shaped post looks like the argument the writer is actually making. Subheads that are claims, not labels. Paragraphs that are different lengths because the thoughts inside them are different sizes. Lists that appear only where a list is the actual form of the thought, not as a scannability hack. The occasional long paragraph because the idea needed room to land. The occasional single sentence because the previous paragraph needed silence after it.

A human-shaped post is also missing things. No “Key Takeaways” panel summarizing the post for the reader who didn’t read it (if they want the TL;DR, they’ll drop your article in an LLM anyway). No FAQ schema designed for a robot. No em dashes scattered like seasoning. The structural restraint is itself a signal: I wrote this for a person, not a crawler.

This is not a step backward in usability. Readers can now feel the difference, even when they can’t name it, because the machine version of the page has become the dominant baseline they are comparing against.

The publishers I respect most are using their CMS the way a sculptor uses a chisel, not the way a stenographer uses a template. They delete the boxes their SEO plugin keeps trying to add. They write a subhead that gives the argument away rather than one optimized for snippet extraction. They allow a paragraph to be eight sentences long if eight sentences are what the thought required.

The trade is real. They lose some ranking. They keep their audience.

Open your last three blog posts and look at the shape, not the words. That’s the article most readers read.