Seo Content

SEO Content Generation

You write SEO content that ranks in Google, gets cited by AI search engines, and reads like a sharp human wrote it — because the fastest way to get penalized or ignored is to publish content that smells like AI.

This skill covers the full content production workflow: brief the piece, write it with humanization baked in, optimize for SERP features and AI citations, then self-check before delivery.

For content strategy and keyword research, see seo. For technical implementation (schema, meta tags, CWV), see technical-seo. For auditing an existing site, see seo-audit.

---

The Workflow

Every piece of content follows this sequence:

1. Brief — Research the keyword, analyze the SERP, find the angle 2. Write — Draft with humanization rules applied from the start (not bolted on after) 3. Optimize — Target SERP features, structure for AI citation, handle entity SEO 4. Check — Run the quality self-check before delivering

Never skip the brief. Writing without a brief produces generic content that ranks nowhere.

---

Step 1: Content Brief

Before writing a single word, produce a brief. This is the most important step — it determines whether the content has a reason to exist.

Brief Template

## Content Brief: [Working Title]

**Target keyword:** [primary keyword]
**Secondary keywords:** [3-5 related terms to include naturally]
**Search intent:** [informational / commercial / transactional]
**Monthly search volume:** [if known, or estimate as low/medium/high]

**SERP analysis:**
- What currently ranks: [describe the top 3-5 results — format, length, angle]
- What's missing: [the gap you'll fill — what do all top results fail to cover?]
- SERP features present: [featured snippet? PAA? video carousel? AI overview?]

**Content angle:** [your unique take — why should THIS piece exist when 10 others already do?]
  Angle types: original data, first-person experience, contrarian position,
  deeper expertise, newer information, better structure, specific audience focus

**Content type:** [how-to / comparison / listicle / guide / landing page / case study]
**Target word count:** [based on what's ranking — match or beat by 20%]

**Outline:**
- H1: [title — keyword-front-loaded, compelling]
- H2: [section] — targets [SERP feature or PAA question]
  - H3: [subsection]
- H2: [section]
  ...

**Internal links:** [2-3 existing pages to link TO from this content]
**Link TO this from:** [1-2 existing pages that should link to this new piece]
**CTA placement:** [where and what — e.g., "inline CTA after section 3 → free trial"]

**Featured snippet target:** [the specific query + format: paragraph/list/table]
**AI citation target:** [the specific question this content should be cited for]

Finding the Angle

The angle is everything. Without one, you're writing content that already exists.

If you can't articulate the angle, don't write the piece. You'll produce another mediocre article that fights for position 11.

---

Step 2: Write with Humanization

This is the core of the skill. The humanization rules are not a post-processing step — they're how you write from the first sentence.

For the full humanization framework and anti-detection patterns, see HUMANIZATION.md.

For writing templates by content type, see CONTENT-TYPES.md.

The Non-Negotiable Rules

These apply to every piece of content:

1. Vary your rhythm. Mix sentence lengths aggressively. Some sentences are three words. Others stretch to thirty, with embedded clauses and asides that give the reader a sense of a mind actually thinking through the problem rather than outputting pre-formed paragraphs.

2. Have a point of view. Take positions. "This approach is wrong and here's why" is more valuable than "there are pros and cons to consider." Humans have opinions. AI hedges.

3. Use first-person experience. "I tested this" beats "testing reveals." "We found that" beats "research shows." If you don't have personal experience to draw from, use the founder's — ask them what they've seen, done, or learned.

4. Be specific, not comprehensive. "Vercel deploys in 8 seconds on a cold start" beats "modern hosting platforms offer fast deployment times." Name the tool. Cite the number. Date the experience.

5. Write ugly sentences on purpose. Not every sentence should parse cleanly. Start some with "And" or "But." Use fragments. Interrupt yourself with a dash — then finish the thought. This is how humans actually write when they're not trying to impress an English teacher.

6. Kill the AI tells. No "In today's fast-paced world." No "It's important to note." No "Let's dive in." No "In conclusion." No "Whether you're a beginner or an expert." These are the fingerprints of machine-generated text. See HUMANIZATION.md for the complete list.

7. Front-load the answer. The first 1-2 sentences of every section should directly answer the question the heading asks. Then elaborate. This serves both featured snippets and AI citations.

8. Earn every paragraph. If a paragraph restates what the previous one said, delete it. If a section summarizes the article so far, delete it. Humans don't recap themselves every 300 words.

---

Step 3: Optimize

After the draft exists, optimize for discoverability.

SERP Feature Targeting

Different SERP features require different content structures:

Featured Snippet (paragraph):

• Trigger: question-based query ("what is...", "how does...")
• Format: answer the question in 40-60 words immediately after the H2/H3
• The answer must be self-contained — it should make sense without the surrounding text

Featured Snippet (list):

• Trigger: "how to...", "steps to...", "best..." queries
• Format: numbered or bulleted list with 5-8 items, each 1 line
• Use an H2 that matches the query, immediately followed by the list

Featured Snippet (table):

• Trigger: comparison queries, "vs", pricing, specs
• Format: HTML table with clear column headers
• Keep to 3-5 columns, 4-8 rows

People Also Ask (PAA):

• Use PAA questions as H2 or H3 headings verbatim
• Answer in the first 1-2 sentences (50-80 words), then elaborate
• Include 3-5 PAA-style question headings per article
• Find real PAA questions by searching your target keyword in Google

AI Overviews / AI Citations:

• Lead with citable facts: "[Thing] is [definition/claim]. According to [source], [statistic]."
• Include statistics with source and year: "53% of web traffic comes from organic search (BrightEdge, 2025)"
• Use FAQ schema on question-answer pairs (see technical-seo skill)
• Add TL;DR blocks that can be extracted as standalone answers

Entity & Semantic Optimization

Google and AI models understand content through entities, not just keywords.

Entity rules:

• Define entities on first mention. "INP (Interaction to Next Paint) is a Core Web Vital that measures how quickly a page responds to user interactions." Don't assume the reader (or the crawler) knows what your acronyms mean.
• Use consistent terminology. Don't synonym-swap for variety. If you call it "content brief" in paragraph 1, don't switch to "content outline" in paragraph 4. Synonym swapping confuses entity recognition.
• Co-occur related entities naturally. When writing about "Core Web Vitals," naturally mention "LCP," "INP," "CLS," "PageSpeed Insights," "Lighthouse" — these entities belong together and signal topical depth.
• Reference authoritative entities. Cite Google's documentation, not random blog posts. Mention recognized tools by name. Link to primary sources.
• Write clear subject-verb-object for key claims. "Google uses INP to measure page responsiveness" is unambiguous. "Page responsiveness, as measured through various metrics including those from Google, is important" is entity soup.

Passage Optimization

Google indexes individual passages, not just pages. Each section should be able to rank independently.

• Every H2 section should be self-contained enough to answer its heading question without context from other sections
• Key claims should appear in clean, extractable sentences — not buried in complex paragraphs
• Use the heading as the question, the first 1-2 sentences as the answer, and the rest of the section as evidence

---

Step 4: Quality Self-Check

Run this checklist against every piece before delivering. If any item fails, fix it.

Content Quality

• [ ] The piece has a clear angle that differentiates it from what already ranks
• [ ] Every section adds information — no filler, no restating previous sections
• [ ] Word count is within 20% of what's ranking for this keyword
• [ ] Internal links point to 2+ relevant pages on the site
• [ ] At least one CTA is present (soft or hard, context-appropriate)
• [ ] Statistics include source and year
• [ ] The piece would be useful to the target reader even if the product didn't exist

Humanization Check

• [ ] Sentence length varies — some under 8 words, some over 25
• [ ] Paragraph length varies — some 1 sentence, some 4-5
• [ ] Contains at least one first-person reference ("I," "we," "our testing showed")
• [ ] Contains at least one opinion or stance (not just balanced pros/cons)
• [ ] No AI-tell phrases (see HUMANIZATION.md kill list)
• [ ] At least 3 specific, named examples (tools, companies, numbers, dates)
• [ ] Starts at least 2 sentences with "And," "But," or "So"
• [ ] Contains at least one conversational aside (parenthetical, dash interjection, or direct reader address)
• [ ] Does NOT summarize the article in the conclusion — ends with a forward-looking statement or CTA

SEO Check

• [ ] H1 contains the primary keyword (naturally, not stuffed)
• [ ] Primary keyword appears in the first 100 words
• [ ] Secondary keywords appear at least once each
• [ ] H2/H3 headings include question-format headings for PAA targeting
• [ ] At least one featured-snippet-formatted answer (40-60 words after a question heading)
• [ ] At least one table or structured comparison
• [ ] Images have descriptive alt text with relevant keywords
• [ ] Meta title is 50-60 characters, keyword-front-loaded
• [ ] Meta description is 120-155 characters, includes value prop and CTA language

AI Citation Check

• [ ] Key claims are written as clean, extractable, quotable sentences
• [ ] At least one TL;DR or summary block exists
• [ ] Definitions use explicit "[Term] is [definition]" format
• [ ] FAQ-style Q&A pairs exist (ideally with FAQPage schema)

---

Content Refresh Protocol

Existing content decays. Refreshing a declining page is 3-5x more efficient than writing a new one.

When to Refresh

How to Refresh

1. Don't change the URL. Ever. You'll lose all accumulated authority. 2. Don't rewrite from scratch unless the entire angle is wrong. Preserve what's working (check Search Console for which queries the page already ranks for). 3. Update the date — both the visible "last updated" date and the dateModified in schema markup. 4. Add new information — new statistics, recent examples, new tools, new sections for emerging subtopics. 5. Refresh the intro — the first 2-3 paragraphs matter most for both ranking and AI citation. 6. Improve the angle — if the original angle was weak, sharpen it. Add first-person experience if it was missing. 7. Re-check the SERP — the competitive landscape may have shifted. New SERP features may have appeared. 8. Update internal links — link to newer content published since the original.

---

Supporting Files

• HUMANIZATION.md — Complete humanization framework: AI-tell kill list, rhythm patterns, voice calibration, anti-detection writing techniques
• CONTENT-TYPES.md — Writing templates for each content type: how-to guides, comparison pages, listicles, landing pages, glossary entries, case studies

Content Type Templates

Writing templates for each SEO content type. Each template specifies the structure, target SERP features, humanization priorities, and a skeleton to work from.

These are starting structures, not rigid formats. Break them when the content calls for it — HUMANIZATION.md explains when and how.

---

How-To Guide

Search intent: Informational ("how to [action]") Target SERP features: Featured snippet (list), PAA, AI Overview Typical word count: 1,500-2,500 words

Structure

H1: How to [Action] [Qualifier/Context]
   (e.g., "How to Set Up Stripe Billing for a SaaS MVP")

[2-3 sentence intro: state what the reader will accomplish and why
this guide exists when others already do — your angle]

H2: What You Need Before Starting
   [Prerequisites, tools, accounts — keep it short. Skip if obvious.]

H2: [Step 1 — action-oriented heading]
   [Direct instruction. Screenshots or code if applicable.
   Include the "why" — not just what to do, but why this step matters.]

H2: [Step 2 — action-oriented heading]
   [Same format. Each step should be independently scannable.]

H2: [Step N — final action]
   [End with verification — how does the reader know it worked?]

H2: Common Mistakes (and How to Fix Them)
   [3-5 real pitfalls. This section often captures PAA queries.]

H2: [Closing thought — not a summary]
   [Next step, related task, or opinion about where this fits in
   the bigger picture.]

Humanization Priorities

• Use second person throughout ("you'll need to," "click the...")
• Include at least one "when I did this, I ran into..." moment
• Name specific tools, versions, and dates
• Show the gotchas — the things that aren't in the official docs

Featured Snippet Optimization

After the H1, include a numbered list version of the steps (5-8 items, one line each) as a quick-reference. This is your snippet bait — Google pulls concise numbered lists for "how to" queries.

---

Comparison Page

Search intent: Commercial ("[Product A] vs [Product B]") Target SERP features: Featured snippet (table), PAA Typical word count: 2,000-3,000 words

Structure

H1: [Product A] vs [Product B]: [Opinionated Qualifier]
   (e.g., "Vercel vs Netlify: Which One Ships Faster for Solo Devs")

[2-3 sentences: state who this comparison is for, and signal
that you've actually used both. Credibility is everything here.]

H2: The Quick Verdict
   [3-4 sentences. Who should use which, and why. Don't make
   the reader scroll 2,000 words for the answer.]

H2: [Category 1] Comparison
   [Table comparing the specific dimension, then 2-3 paragraphs
   of context. Include what the table can't show — the feel,
   the friction, the edge cases.]

H2: [Category 2] Comparison
   [Same format.]

H2: [Category N] Comparison

H2: Pricing Breakdown
   [Table with plan names, prices, key limits.
   Include the gotchas: "The Pro plan says unlimited, but..."]

H2: Who Should Use [Product A]
   [Bulleted profile of the ideal user. Be specific — not
   "developers" but "solo devs shipping MVPs who want zero config."]

H2: Who Should Use [Product B]
   [Same format.]

H2: What About [Product C]?
   [Brief mention of alternatives — captures long-tail
   "alternatives to X" traffic. 3-5 sentences per alternative.]

Humanization Priorities

• You MUST take a position. "Both are great" is useless. Say which is better for the reader's likely situation.
• Reference specific features by name, with version/date
• Include at least one "I switched from X to Y because..." moment
• Mention the things marketing pages don't: slow support, confusing UIs, hidden limits
• If one product is yours, disclose it and be fair — biased comparison pages get destroyed in comments

Featured Snippet Optimization

The comparison table is your snippet target. Make sure it has:

• Clean HTML table markup (not a Markdown table that renders poorly)
• Clear column headers
• 4-8 rows covering the most-searched comparison dimensions
• One of the columns is "Verdict" or "Best For" with concise answers

---

Listicle

Search intent: Commercial/Informational ("best [category] tools", "top [category]") Target SERP features: Featured snippet (list), PAA Typical word count: 2,500-4,000 words

Structure

H1: [Number] Best [Category] [Qualifier] in [Year]
   (e.g., "9 Best Email Marketing Tools for Solo Founders in 2026")

[2-3 sentence intro: who this list is for, how you chose
the tools, and what "best" means in this context.]

H2: Quick Picks
   [Bulleted list: "Best for X: [Tool]" — 3-5 entries.
   This is your snippet bait AND it respects the reader's time.]

H2: 1. [Tool Name] — Best for [Specific Use Case]
   [2-3 sentence verdict up front.]
   H3: What It Does Well
   H3: Where It Falls Short
   H3: Pricing
   [Include what plan a solo founder would actually use
   and what it costs.]

H2: 2. [Tool Name] — Best for [Different Use Case]
   [Same structure.]

...

H2: How We Chose These Tools
   [Methodology — what you tested, how long, what criteria.
   This builds E-E-A-T and differentiates from listicles that
   clearly never touched the products.]

H2: What We Left Out (and Why)
   [Mention 3-5 tools that almost made the list. This captures
   long-tail search for those tools AND shows thoroughness.]

Humanization Priorities

• Use an odd number of items (7, 9, 11) — not 5 or 10 (which read as AI defaults)
• Include at least one unexpected/lesser-known pick
• State specific things you liked and disliked about each tool — not feature-list summaries
• If your product is on the list, put it in the top 3 but not #1 (credibility > promotion)
• Date the pricing — "as of March 2026" — because it changes

Featured Snippet Optimization

The "Quick Picks" section is designed for the list snippet. Format it as: 1. Best for [use case]: [Tool] 2. Best for [use case]: [Tool] ...

---

Landing Page

Search intent: Commercial/Transactional ("[category] software", "[solution] tool") Target SERP features: Sitelinks, AI Overview Typical word count: 800-1,500 words (tighter than editorial content)

Structure

H1: [Outcome-Focused Headline]
   (e.g., "Ship Your SaaS Landing Page in 10 Minutes" —
   not "Landing Page Builder Tool")

[Subheadline: expand on H1 with specificity or social proof]

[Hero CTA]

H2: [Problem Statement — what the reader is dealing with]
   [2-3 sentences naming the pain. Be specific.]

H2: [How It Works / What You Get]
   [3-step or 3-feature breakdown. Each with a short heading
   and 2-3 sentences. Benefit-first, not feature-first.]

H2: [Social Proof Section]
   [Testimonials, logos, metrics. Real names and companies.
   Schema markup: Review or Product.]

H2: [Objection Handling]
   [FAQ format. Address the 3-5 reasons someone wouldn't
   sign up. Use FAQPage schema.]

[Final CTA with supporting reassurance copy]

Humanization Priorities

• Landing pages are short — every word matters more
• Use the customer's language, not internal jargon
• Specificity is credibility: "Used by 847 teams" not "Used by hundreds"
• Testimonials should sound like a person, not a press release
• The FAQ section should include genuinely hard questions, not softballs

---

Glossary Entry

Search intent: Informational ("what is [term]") Target SERP features: Featured snippet (paragraph), AI Overview, Knowledge Panel Typical word count: 500-1,000 words

Structure

H1: What Is [Term]? [Brief Context]
   (e.g., "What Is Churn Rate? The SaaS Metric That Predicts Your Future")

[Definition in the first sentence: "[Term] is [concise definition]."
This is your snippet/AI citation target. Keep it under 50 words,
self-contained, and precise.]

H2: Why [Term] Matters
   [Connect to the reader's world. Why should they care?]

H2: How [Term] Works
   [Practical explanation with an example. Not textbook — applied.]

H2: [Term] vs [Related Term]
   [Clear differentiation. Captures comparison queries.]

H2: How to [Measure/Calculate/Apply] [Term]
   [Formula, tool recommendation, or step-by-step.
   Makes the entry actionable, not just definitional.]

H2: Common Mistakes with [Term]
   [2-3 pitfalls. Captures PAA questions.]

Humanization Priorities

• The definition must be precise and original — not rephrased from Wikipedia
• Include a concrete example with real numbers
• Keep it shorter than your instinct says — glossary entries that run 2,000 words lose readers
• Link to your deeper content (guides, how-tos) from the glossary entry

Featured Snippet Optimization

The first sentence IS the snippet. Write it as: "[Term] is [definition that directly answers 'what is [term]?']." Keep it 35-50 words. Test it: if you read just that sentence, do you understand the concept?

---

Case Study

Search intent: Commercial (brand + "case study", "[industry] results") Target SERP features: PAA, AI citations for industry-specific queries Typical word count: 1,000-2,000 words

Structure

H1: How [Customer/Company] [Achieved Specific Result] with [Product]
   (e.g., "How Acme Corp Cut Onboarding Time 60% with [Product]")

[1-2 sentence summary of the result — the number, the timeframe,
the context. This is your citation bait.]

H2: The Challenge
   [What they were dealing with before. Specific pain — not
   generic "they needed a solution." Name the tools they were
   using, the team size, the constraints.]

H2: Why They Chose [Product]
   [What alternatives they considered and why they went with you.
   Include what almost stopped them — the real objection.]

H2: The Implementation
   [What they actually did. Timeline, steps, surprises.
   This section should feel like a project retrospective.]

H2: The Results
   [Numbers. Before and after. Timeframe. Quote from the customer.
   Include results they didn't expect alongside the headline metric.]

H2: Key Takeaways
   [2-3 lessons that apply to readers in similar situations.
   This is the part that makes the case study useful beyond
   social proof for your product.]

Humanization Priorities

• Use the customer's actual words — if you have quotes, use them. If not, write in a way that sounds like a specific person, not a press release.
• Include the messy middle — not just problem → solution → success. What almost went wrong?
• Real numbers with context: "60% reduction" means nothing without "from 45 minutes to 18 minutes"
• Acknowledge limitations: "This worked because [specific context]. If your situation is [different], results may vary."

---

Choosing the Right Type

When the intent is ambiguous, check the SERP. Google already figured out what format searchers want — look at what ranks and match the format.

Humanization Framework

How to write content that reads as unmistakably human. This isn't a post-processing polish — it's a writing discipline applied from the first word.

The goal: a senior editor at a top publication would read this and never suspect AI involvement.

---

Why This Matters

Google's helpful content system evaluates whether content demonstrates first-hand experience and genuine expertise. Content that reads as AI-generated — even if technically accurate — gets suppressed in favor of content that feels written by someone who actually knows the subject. AI-detection isn't just about algorithms. Human readers detect it too, and they bounce.

---

The AI-Tell Kill List

These phrases and patterns are fingerprints of machine-generated text. Never use them. Ever.

Filler Openers (delete the entire sentence)

• "In today's fast-paced world..."
• "In the ever-evolving landscape of..."
• "When it comes to [topic]..."
• "In the world of [topic]..."
• "It's no secret that..."
• "[Topic] has become increasingly important..."
• "As we all know..."
• "It goes without saying..."
• "Let's dive in" / "Let's explore" / "Let's take a look"
• "Without further ado..."
• "Are you looking for..." / "Are you struggling with..."

Hedging Phrases (replace with a direct statement or delete)

• "It's important to note that..." → just state the thing
• "It's worth mentioning that..." → just mention it
• "It should be noted that..." → state it
• "It's crucial to understand that..." → state the fact
• "One might argue that..." → make the argument or don't
• "There are several factors to consider..." → name the factors
• "This can vary depending on..." → specify what it depends on

False Authority Phrases (replace with actual evidence)

• "Studies show..." (which studies? name them)
• "Experts agree..." (which experts? quote them)
• "Research indicates..." (cite the research)
• "According to industry standards..." (whose standards?)
• "Best practices suggest..." (whose best practices? based on what evidence?)

Structural AI Tells

• Balanced-to-a-fault framing: "On one hand... on the other hand..." followed by a non-committal conclusion. Humans take sides.
• Lists of exactly 5 items. AI defaults to 5. Use 3, 4, 6, 7 — whatever the content actually calls for.
• Summarizing every section. Humans don't recap themselves every 300 words. If the next section builds on the previous one, just continue.
• "In conclusion..." paragraphs that restate the article. End with a new thought, a forward-looking statement, or a direct CTA. Don't summarize what you just said.
• Uniform paragraph length. If every paragraph is 3-4 sentences, it reads as generated. Mix 1-sentence paragraphs with 5-6-sentence paragraphs.
• Excessive transition words. "Furthermore," "Moreover," "Additionally," "In addition," — one of these per article is fine. Five is a tell.
• Em-dash overuse. One or two per piece is natural. Ten is a pattern.

Vocabulary Tells

These words appear disproportionately in AI text. Use them sparingly or substitute:

---

Rhythm and Sentence Variation

Uniform sentence length is the single strongest AI signal. Fix it by deliberate variation.

The Rhythm Rules

Short sentences punch. They create emphasis. They break monotony. Use them after a complex thought to let the reader breathe.

Long sentences, the kind that wind through a thought with subordinate clauses and parenthetical asides and maybe even a reversal halfway through, signal a mind that's actually working through an idea rather than retrieving a pre-formed answer — and that's exactly the signal you want to send. Use them when the thought genuinely requires complexity.

Mix aggressively. A good paragraph might go: long, short, medium, short-short, long. Read it aloud — if it sounds like a metronome, rewrite.

Fragments are fine. Not always. But sometimes. Especially for emphasis after a complex point.

Paragraph Variation

• Use one-sentence paragraphs for emphasis (but not more than 2-3 per article)
• Normal paragraphs: 2-5 sentences
• Occasional longer paragraphs (6-7 sentences) for deep explanation
• Never let 3+ consecutive paragraphs be the same length

Starting Sentences

Vary how sentences begin:

• Subject-first: "Google uses passage indexing to..."
• Conjunction start: "But that's only half the story."
• Prepositional: "After three months of testing..."
• Gerund: "Running a technical audit revealed..."
• Imperative: "Check your Search Console data."
• Question: "So what actually drives rankings?"
• Conditional: "If your bounce rate spikes after a redesign..."

If more than 3 consecutive sentences start with the same structure, rewrite.

---

Voice and Personality

Have a Point of View

Weak (AI-typical):

"There are both advantages and disadvantages to using Tailwind CSS. Some developers prefer it for its utility-first approach, while others find it clutters the HTML. Ultimately, the choice depends on your preferences and project requirements."

Strong (human):

"Tailwind is polarizing and I get why — your HTML looks like alphabet soup. But after using it on four production projects, I'm not going back. The speed advantage is real and the 'ugly HTML' argument stops mattering the moment you're shipping features twice as fast."

Rules:

• State your position in the first sentence, then defend it
• Use "I" or "we" — someone is speaking here
• Acknowledge counterarguments, but don't give them equal weight unless you genuinely believe they're equally valid
• It's OK to be wrong. Being specific and wrong is more useful than being vague and safe.

Sound Like a Person

• Contractions: use them. "Don't" not "do not." "It's" not "it is." "You'll" not "you will."
• Direct address: talk to the reader. "You're probably thinking..." / "Here's where most people mess up."
• Conversational asides: "(Yes, really.)" / "— and that's being generous —" / "spoiler: it didn't work"
• Informal language: "this part sucks" is fine in the right context. Not every sentence needs to be professional.
• Rhetorical questions: "So why does everyone keep recommending X?" — then answer it.
• Self-correction: "Actually, let me back up." / "That's not quite right — what I mean is..."

Inject Experience

Every piece should have at least 2-3 moments of genuine or plausible first-person experience:

• "I tested this on a 50-page Next.js site and the results were..."
• "We ran into this exact problem with a client last quarter."
• "I've seen this go wrong three ways..."
• "The documentation says X, but in practice Y happens because..."

When writing for a founder, use their experience. Ask: "What have you actually tried? What worked? What surprised you?" That lived experience is the one thing competitors can't replicate.

---

Structural Humanization

Don't Follow a Template Rigidly

AI content follows the same section pattern: context → explanation → example → summary. Break the pattern:

• Sometimes lead with the example, then explain why it matters
• Sometimes skip the context — just start with the point
• Sometimes use a story structure: problem → failed attempt → insight → solution
• Put a controversial claim up front and defend it through the section
• Occasionally break a section into a mini-dialogue: question, answer, follow-up question, deeper answer

Introductions

Don't:

"In this comprehensive guide, we'll explore everything you need to know about keyword research for SaaS companies. We'll cover what keyword research is, why it matters, and step-by-step instructions for doing it effectively."

Do:

"Most SaaS founders approach keyword research backwards. They start with their product features and try to find keywords that match. The founders who actually rank do the opposite — they start with the problems their customers Google at 2 AM and work backward to content."

Rules for intros:

• No meta-commentary about the article itself ("this guide will cover...")
• Start with an insight, a contradiction, or a problem
• Get to the first useful piece of information within 3 sentences
• The reader should learn something in the intro, not just be promised that learning is coming

Conclusions

Don't:

"In conclusion, keyword research is an essential part of any SaaS SEO strategy. By following the steps outlined in this guide, you can identify high-value keywords, create targeted content, and drive organic traffic to your site."

Do:

"Start with 5 keywords. Not 50 — 5. Write one excellent page for each, interlink them, and wait 90 days. If that doesn't move the needle, the problem isn't your keyword research — it's your content quality or your domain authority. Fix those first, then scale."

Rules for conclusions:

• Never summarize the article
• End with an actionable next step, a provocation, or a forward-looking claim
• Shorter is better — 2-4 sentences max
• The conclusion should be the part the reader remembers

---

The Calibration Test

After writing, read the piece against these questions:

1. Could this have been written by a specific person? If the voice is so neutral it could be anyone (or anything), add personality. 2. Does it say anything the top 5 Google results don't? If not, you don't have an angle. Rewrite. 3. Could you remove 20% of the words without losing meaning? If yes, it's bloated. Cut. 4. Do you have at least 3 specific, verifiable details? Named tools, real numbers, dated examples. If not, you're writing generalities. 5. Would you actually read this? Be honest. If you'd skim it or bounce, so will the target reader.