How a Dan Brown Style File Beat Google's AI Content Update

Short answer: When Google's update targeting AI-generated content sank impressions on a 50-post site, I recovered by having an AI analyze Dan Brown's prose and distilling it into a reusable writing style skill file. The posts stopped reading like "AI content" — and impressions climbed to nearly double the previous peak. This post shows you, step by step, how to build the same file for yourself.

The Backstory: Why I Needed This File

If you've read this series from the start, you'll remember the orhunercan.com case and its chart from the GEO guide. The short version:

The site had about 50 blog posts produced with AI assistance, and the first SEO push had paid off on April 23. Then, in mid-May, Google shipped an update targeting low-quality AI content. Impressions went into decline.

The problem wasn't the content's information — it was its language. The posts were accurate and useful, but they all read like they came out of the same mold. That familiar AI rhythm: evenly sized sentences, "in conclusion" endings, paragraphs that never take a risk.

The fix wasn't rewriting the content. It was intervening in the language layer of the production pipeline.

What Is a Writing Style Skill File?

A writing style skill file is an author's prose — sentence rhythm, paragraph construction, narrative technique — analyzed by AI and distilled into a reusable set of rules. You build it once; every piece of content you produce afterwards passes through it.

Here's the key distinction: telling an AI to "write fluently" doesn't work — the output of that instruction is still average AI prose. You have to give it concrete, checkable rules: "no three consecutive sentences of the same length," "every section ends on an unanswered question." The skill file is the repository of exactly those rules.

Why It Works — With Data

This approach working isn't luck; there are two independent lines of evidence:

1. GEO research: The Princeton-led GEO study measured that among nine tested optimization methods, fluency optimization boosted visibility by 15–30%. A writing style skill file is fluency optimization, systematized.
2. Google's own framework: Google states officially that the criterion is usefulness to the user, not how content was produced. What's being targeted isn't AI — it's uniform, careless production. Content that breaks the mold survives, even AI-assisted — the orhunercan.com chart is the proof.

There's also the invisible side: reader behavior. Compelling prose gets read longer, and longer dwell time with lower bounce rates are indirect but real ranking signals.

Step by Step: Build Your Own Style Skill File

1. Pick your model author

I used three criteria:

• Proven mass audience: The author should be read by millions, not just literary circles. The goal is readability, not art.
• Pull: Prose that makes you turn the page. The blog equivalent: prose that makes you keep scrolling.
• Firsthand familiarity: During my military service I read stacks of Dan Brown novels to unwind — I knew from experience how that prose drags you in. That familiarity becomes your quality control when you evaluate the AI's analysis.

Dan Brown checked all three boxes. Your author may differ; the criteria shouldn't.

2. Collect sample passages

Pick passages serving different functions, from different books: scene openings, tension build-ups, information-heavy passages, chapter endings. Feed it one type of sample and the analysis comes back one-dimensional. The goal is to extract the author's toolbox, not a single trick.

3. Have the AI analyze them

Give the passages to an AI and request analysis along these axes:

• Sentence rhythm: What's the length distribution? When does the short sentence arrive? How does rhythm change at moments of tension?
• Paragraph construction: How many sentences? How do paragraphs open and close? What are single-sentence paragraphs used for?
• Narrative technique: How is information staged? How is curiosity created, and how long is it held open? How is the reader addressed?
• Transitions: How are sections and topics connected?

4. Distill the analysis into rules

The analysis output will be long and literary — you need to distill it into enforceable rules. The skeleton of my file looks like this:

# Writing Style Guide

## Sentence Rhythm
- No three consecutive sentences of the same length.
- The short sentence (3-5 words) is an emphasis tool; it lands right after critical information.
- The long sentence sets the scene; then a short one cuts it.

## Paragraph Construction
- Paragraphs run 2-4 sentences. A single-sentence paragraph = the strongest emphasis.
- Every paragraph ends leaving a small hook that pulls into the next.
- Section endings: an unanswered question or an unexpected fact.

## Narrative
- Deliver the abstract concept through a concrete scene first; explain after.
- Address the reader directly ("imagine this", "if it were you...").
- A technical term is defined in the sentence where it first appears, not deferred to a glossary.

## Forbidden
- Cliché openers in the "In today's digital world..." family.
- Closings that start with "in conclusion" or "to summarize".
- Passive voice in two consecutive sentences.
- The same connective twice in one paragraph.

This is a skeleton — the real file is longer and author-specific. But note the format: every rule is checkable. It doesn't say "be fluent"; it says "no three consecutive sentences of the same length." That's the kind of rule an AI can actually apply.

5. Wire it into the production pipeline

A file sitting on a shelf does nothing. Instead of theory, here's the actual workflow I built while recovering those 50 posts:

1. I connected my AI to Google Search Console via MCP. (MCP — Model Context Protocol — is the open standard that lets an AI assistant connect directly to external tools and data.) Then I had the AI itself diagnose which pages the decline was coming from: which post lost how many impressions, one by one.
2. I connected Google Ads the same way and ran the keyword research from there: which queries carry volume, which keywords hold opportunity.
3. I had two skill files ready: an SEO skill file (technical and content rules) and the Dan Brown writing style skill file (prose rules).
4. Then I gave a single instruction: "Revise my blog posts according to this data and in light of these two skill files."

The AI combined the Search Console diagnosis ("which pages are falling"), the keyword opportunities from Ads, and the rules in both files — and reworked the posts. My job wasn't writing line by line; it was running the pipeline and auditing the output.

The power of this setup: data (what to fix), strategy (which keyword to target), and style (how to write it) meet in the same production line. On its own, a skill file is decoration; inside this pipeline, it's a multiplier.

6. Measure the result

There's one scoreboard: Search Console. Annotate the date of the change, then watch the impressions and clicks curves weekly. In my case the recovery started within weeks, and the curve climbed to nearly double the previous peak — the full chart is in the first post of this series.

In fairness: the skill file wasn't the only variable in that period; a multi-layered SEO effort was running alongside it. But I believe the move that separated me from competitors was this file — because everyone does technical SEO, and nobody was producing AI content that reads like a human.

Ethics and Limits

Three lines worth drawing clearly:

• This is not plagiarism. The file learns how the author builds sentences, not the sentences themselves. Style is not subject to copyright; the topic, information, and claims of your text are entirely your own.
• It doesn't substitute for content quality. Writing wrong information in Dan Brown's rhythm doesn't make it right. The skill file is a delivery layer, not packaging — the substance comes from you.
• The bar is higher in sensitive niches. In YMYL areas like health and finance, style alone is not enough; expert review and citations are mandatory. On orhunercan.com, every piece passes physician review.

Frequently Asked Questions

What is a writing style skill file? It's an author's prose — sentence rhythm, paragraph construction, narrative technique — analyzed by AI and distilled into a reusable set of writing rules. The rules are fed to the AI on every content run, pulling the output out of the generic "AI content" mold.

Is this method plagiarism? No. The skill file learns the author's style — rhythm, structure, technique — not their content. Style is not subject to copyright; the topic, information, and claims of every text you produce are entirely your own.

Does Google penalize AI content? Officially, Google targets low-quality, mass-produced content rather than AI use itself; the stated criterion is usefulness to the user, not how content was produced. In practice, text carrying formulaic AI language loses rankings.

Why Dan Brown? Because his books reach some of the widest audiences in the world and their pull is proven. The same method works with any author who writes for a broad audience and whose prose you know well.

Closing

For most sites, that Google update was a penalty; for me, it was a forced-innovation moment. The lesson: producing content with AI isn't the problem — producing content that reads like AI is. What closes the gap is a systematic style layer you build once and apply on every run.

In the next post in this series, I'll start sharing GrowthMadre's own AI visibility data — the results of the GEO layer we applied to this site, with the numbers.