# Usage: curl -sSL https://seed.show/marketing.seo.strategy | bash -s <install-path>
# <install-path> is the directory where the file should land.

set -euo pipefail
[ -z "${1:-}" ] && {
  echo "install requires a path: curl -sSL https://seed.show/marketing.seo.strategy | bash -s <install-path>" >&2
  exit 1
}
TARGET="$1"
mkdir -p "$TARGET"
DEST="$TARGET/seed-fold.PrXpUg.folded.md"

cat > "$DEST" <<'PORTDOWN_24A5A142'
<!--
This is a `.folded.md` archive — a directory packed into one markdown
file. The markers below are load-bearing; don't edit them directly.

To unpack (requires bash — if you have no shell, tell the user):
  1. curl -sSLf https://fold.dom.vin/skill | bash -s <INSTALL_DIR>
  2. <INSTALL_DIR>/fold/scripts/unfold <this-file>
     (or: unfold <this-file>  if fold/scripts is on your PATH)
-->

---
fold: true
marker: 2da7ad
at: 2026-05-07T16:16:18Z
root: seed-pack.EBM5ih
---

<!--fold:2da7ad@file path="README.md" mode="644"-->
# marketing.seo.strategy

SEO strategy context for agents. What to know before touching a keyword brief, advising on content architecture, running a technical audit, or measuring organic performance.

**Before proceeding:** Ranking factor weights are not public. Any source claiming to know the exact weight of a signal is guessing or extrapolating from correlation studies. What is public: structural principles, confirmed algorithm changes, and Google's own guidance. This seed covers the stable structural logic; fetch sources.md for authoritative current documentation.

## What SEO is actually trying to do

SEO is not about gaming rankings. It is about building **durable organic traffic** by creating content that genuinely answers searcher intent, earning links because the content is worth citing, and maintaining a technical foundation that search engines can reliably crawl and index.

The distinction matters because every shortcut that treats SEO as a ranking game produces content that either doesn't earn traffic or earns it temporarily and then loses it. Google's core algorithm updates — Panda, Penguin, Helpful Content, core quality updates — have all moved in the same direction: rewarding content written for people, penalizing content written for crawlers. This directional consistency is stable even when the exact signal weights are not.

## The three pillars and when each matters

**Technical foundation.** Crawlability, indexability, site speed, mobile usability, Core Web Vitals. This is the prerequisite layer. A technically broken site can't rank regardless of content quality or links. For new sites and large sites undergoing migration, technical work comes first. For established sites with stable traffic, technical issues are usually marginal improvements, not the primary lever.

**Content relevance.** Pages that match search intent, cover the topic with appropriate depth, and demonstrate expertise. This is the primary driver for most sites. Content relevance determines whether a page can rank for a given query at all; technical and authority factors break ties. The mistake is treating content relevance as keyword density or topic coverage checklists — it means the page genuinely serves the person who searched.

**Authority and links.** Backlinks from relevant, authoritative sites signal that the content is worth citing. Links remain a significant ranking factor, particularly for competitive queries where multiple high-quality pages exist. For new sites, early link building is often rate-limiting. For established sites, content quality drives link acquisition naturally; active link building is most valuable for launching new content or entering new topic areas.

**Relative weight by site maturity:**
- New site (0–12 months): Technical foundation first, then content relevance. Links are hard to earn at scale early; focus on being technically sound and covering a coherent topic cluster deeply.
- Growing site (1–3 years): Content relevance is the primary lever. Build topical authority by covering a domain completely, not scattered keywords. Technical issues worth fixing are usually crawl budget and Core Web Vitals.
- Established site (3+ years): Authority is the differentiator in competitive queries. Link building strategy matters. Technical work is maintenance, not transformation.

## Search intent taxonomy

Every query expresses one of four intents. Getting this wrong is the most common content architecture failure.

**Informational.** The searcher wants to learn something. "How does a 401k work." "What is quantum computing." "Why is my plant dying." Content type: educational articles, guides, explainers. These attract readers early in a decision journey; they're highest volume, lowest direct conversion, and best for building topical authority and earning links.

**Navigational.** The searcher is looking for a specific thing they know exists. "Stripe pricing page." "Claude API docs." "HubSpot login." Content type: the destination itself — product pages, docs, login pages. These queries are brand-owned; you rank for your own navigational queries or you have a brand awareness problem, not an SEO problem.

**Commercial investigation.** The searcher is comparing options before a decision. "Best project management tools for remote teams." "Notion vs Coda." "Ahrefs review." Content type: comparison pages, roundups, detailed reviews. These sit at the bottom of the funnel; conversion intent is high but not immediate.

**Transactional.** The searcher is ready to act. "Buy running shoes." "Sign up for Figma." "Download QuickBooks." Content type: product pages, landing pages, signup flows. Highest commercial value; these compete with ads on many queries.

**Why intent determines content type:** A page written as a comparison guide will not rank for a transactional query, and a product page will not rank for an informational query. Search engines infer intent from the query and match it to page type. Creating "SEO content" without classifying intent first produces pages that are technically about the topic but structurally wrong for the query.

## What agents get wrong

**Keyword stuffing.** Inserting a target keyword phrase into a page at high density does not improve rankings in modern search. It degrades readability, signals low-quality content, and can trigger manual or algorithmic penalties. Keyword research informs what topic to cover and how the searcher phrases their query; it's not a production spec for word frequency.

**Ignoring search intent.** Targeting a keyword without asking what the searcher actually wants produces content that ranks briefly (if at all) and then fails. A how-to guide targeting a transactional keyword won't convert. A product page targeting an informational keyword won't rank. Intent classification is not optional — it determines whether the content can succeed at all.

**Treating all links as equal.** A link from a topically relevant, high-authority site that people actually read is worth far more than a link from a directory or a blog network. Link building tactics that optimize for link volume rather than link quality produce a backlink profile that looks like spam and performs like spam. Relevant links from real publications matter; everything else is noise at best, risk at worst.

**Exact-match anchor text optimization.** Over-optimizing anchor text so backlinks use the exact keyword phrase you're targeting is a pattern Google's Penguin update was specifically designed to penalize. Natural backlink profiles have varied anchor text. Exact-match anchor text at scale is a spam signal.

**Over-indexing on search volume vs. relevance.** High-volume keywords are competitive; ranking for them requires substantial authority that most sites don't have. Low-volume, high-intent, topically relevant keywords frequently convert better and are achievable. An agent that filters keyword research to "only show me high-volume terms" is optimizing for a vanity metric.

**PageRank as a proxy for quality.** Domain Authority (Moz) and Domain Rating (Ahrefs) are third-party metrics that approximate link-based authority. They correlate with rankings but are not Google signals. Treating DA/DR as a target metric can lead to buying links or chasing placements that look authoritative by these metrics but don't reflect actual quality or topical relevance.

**Treating SEO as separate from UX.** Core Web Vitals (Largest Contentful Paint, Cumulative Layout Shift, Interaction to Next Paint) are direct ranking signals. Page experience — does the page load fast, does it work on mobile, does it not shift around while loading — is part of SEO, not a separate concern. A technically optimized page that provides a poor user experience will underperform a slower page with better UX on competitive queries.

**AI-generated content failure modes.** AI-generated content fails at SEO specifically because it defaults to: covering topics at the same depth as every other page (producing content that matches competitors exactly rather than differentiating), generating text that reads like a topic overview rather than an answer to a specific searcher intent, manufacturing fake expertise signals (vague attributions, hedged claims), and producing content optimized for search engine visibility but not for what the person searching actually needs. The Helpful Content system is explicitly designed to demote content written primarily to rank rather than to serve the reader. AI-assisted content that passes the E-E-A-T test is: written by someone with actual experience with the topic, reviewed by a subject matter expert, specific rather than general, and demonstrably written for a reader, not a crawler.

## What AI is changing

**AI Overviews and zero-click search.** Google's AI Overviews (formerly SGE) generates synthesized answers directly in search results for informational queries. This reduces click-through rates for queries where the AI answer is sufficient. Impact is highest on: simple factual queries, definitions, how-to queries with straightforward answers. Impact is lower on: queries requiring current information, product comparisons, local search, queries with strong navigational intent, and topics where trust and specificity matter. Adapt by optimizing for *featured in AI Overviews* (structured, authoritative content gets cited) and by targeting queries where the answer isn't complete without clicking through.

**AI-generated content and E-E-A-T.** Google's E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) was updated to add "Experience" specifically as AI content scaled. Experience signals that a human with first-hand knowledge wrote or substantially contributed to the content — personal anecdotes, original research, specific examples from practice. AI-generated content that lacks these signals is identifiable and subject to the Helpful Content system's quality assessment. The problem isn't that content was AI-assisted; the problem is that it reads like it was written without any actual knowledge of the topic.

**Automated content farms vs. quality AI-assisted content.** The distinction Google's Helpful Content system draws is: content that exists primarily to rank (regardless of production method) vs. content that exists to serve a reader. AI content farms that produce thin topic overviews at scale — "what is X," "X vs Y," "best X for Y" — without differentiated perspective, original data, or real expertise fail this test regardless of volume. AI-assisted content that substantially improves a subject-matter expert's draft, accelerates research, or helps structure a piece that a human with real experience owns passes it. The dividing line is whether a human with actual expertise was in the loop.

**Helpful content and AI SEO copy.** AI-optimized meta titles, descriptions, and header structures are legitimate uses of AI in SEO workflows. The failure mode is AI-generated body content that substitutes for real expertise. A technically sound page with AI-written body content that demonstrates no actual experience with the topic will underperform a slightly less optimized page written by someone who knows the subject.
<!--fold:2da7ad@file path="mental-models.md" mode="644"-->
# mental-models.md

Key mental models for SEO. These are structural frameworks, not formulas. Apply them to diagnose situations and reason about tradeoffs; don't treat them as checklists.

## Search intent taxonomy with examples

The four intent categories map to distinct content types. The structural match — intent to content type — matters more than keyword density, heading structure, or word count.

| Intent | Signal phrases | Correct content type | Wrong content type |
|---|---|---|---|
| Informational | "how," "what is," "why," "guide to," "explained" | Educational article, tutorial, explainer | Product page, landing page |
| Navigational | Brand name + destination ("login," "pricing," "docs") | The destination itself | Comparison article, blog post |
| Commercial investigation | "best," "vs," "review," "alternatives," "top X for Y" | Comparison, roundup, detailed review | Transactional landing page |
| Transactional | "buy," "download," "sign up," "get started," pricing queries | Product page, signup flow, pricing page | How-to article |

**Diagnosing intent mismatches:** If a page with strong content and good links isn't ranking, the first check is intent match. Look at what Google currently ranks in positions 1–5 for that query. If all ranking pages are a different content type, you have an intent mismatch. The solution is a different page, not more optimization of the existing one.

**Mixed intent queries:** Some queries carry two intents simultaneously. "Best noise canceling headphones" is commercial investigation (comparison), but also informational (help me understand the tradeoffs). "Python tutorial" is informational, but someone searching it might also want to know what to learn after the tutorial (commercial). Read the SERP to understand which intent dominates.

## The content → authority → trust flywheel

The mechanism by which SEO compounds over time:

1. **Create genuinely useful content** that covers a topic with depth and specificity unavailable elsewhere.
2. **Content earns links** from publications, practitioners, and researchers who find it worth citing.
3. **Links signal topical authority** to search engines, improving rankings.
4. **Better rankings drive more discovery**, which surfaces the content to more people who may link to it.
5. **Topical authority accumulates**: once a site has strong link equity in a domain, new content in that domain ranks faster.

Why this matters for strategy: the flywheel takes 12–18 months to noticeably accelerate for a new site. Early SEO work that doesn't produce immediate traffic is not failing — it's building the inputs to the flywheel. The mistake is abandoning the strategy before the compounding effect kicks in, or disrupting it with tactics (spam links, thin content) that signal low quality.

**Breaking the flywheel:** Anything that makes content less trustworthy or less useful breaks the compounding effect. Unhelpful content updates don't just fail to rank new pages — they demote the entire site. This means low-quality pages drag down the authority of high-quality ones on the same domain.

## Crawl budget management

Crawl budget is the number of URLs Googlebot will crawl on your site within a given time period. For most sites under ~10,000 pages, crawl budget is not a constraint — Google crawls everything. It becomes relevant when:

- The site has millions of URLs (e-commerce with faceted navigation, large news sites, UGC platforms).
- Large portions of the site are thin or duplicate content.
- There are significant numbers of redirects, broken links, or URL parameter permutations that create duplicate content.

**The budget principle:** Every crawl credit spent on a low-value URL (parameterized duplicates, pagination beyond reasonable depth, thin category pages) is a crawl credit not spent on content that deserves indexing. For large sites, crawl budget management is active work: robots.txt, `noindex` directives, canonical tags, and reducing crawlable duplicate surface.

**Canonical tags:** A canonical tag (`<link rel="canonical">`) tells Google which version of a page is the "real" one when duplicates exist. Use them when URL parameters create functionally identical pages (e.g., filtered or sorted e-commerce pages) and when content appears at multiple URLs for legitimate structural reasons.

## Site architecture as information architecture

How a site is structured tells search engines which content is important, how topics relate, and how authority should flow. The architecture signals are:

**Internal linking.** Pages linked from many places in the site receive more PageRank internally. The homepage is almost always the highest-PageRank page because everything links to it. Important content should be reachable in 2–3 clicks from the homepage; content that requires 7 clicks to reach is effectively buried.

**Topic clusters.** A cluster is a pillar page covering a broad topic plus supporting pages covering subtopics. The pillar page links to each supporting page; each supporting page links back. This structure concentrates topical authority in the pillar and signals to Google that the site comprehensively covers the topic.

- Pillar: "Project Management Guide" → ranks for broad, high-competition term
- Cluster pages: "Agile vs Waterfall," "How to Write a Project Charter," "Gantt Chart Templates" → rank for specific, lower-competition terms; each links back to the pillar

**Keyword cannibalization.** When multiple pages on the same site target the same keyword, they compete against each other rather than cooperating. Google has to choose one; it often chooses poorly. The solution is consolidation (merge the pages into one authoritative page), differentiation (reorient each page toward a distinct sub-intent), or canonicalization.

**URL structure.** Clean, hierarchical URLs (`/blog/category/post-title`) help search engines understand site structure. Deeply nested URLs, numeric IDs, or URL parameters that create thousands of unique paths create crawl budget waste and make internal linking harder.

## Core Web Vitals as user experience proxies

Google measures three page experience signals under Core Web Vitals:

**Largest Contentful Paint (LCP):** Time until the largest visible content element loads. Represents perceived load speed. Target: under 2.5 seconds. Common causes of failure: large unoptimized images, render-blocking JavaScript, slow server response times.

**Cumulative Layout Shift (CLS):** Visual stability — how much page elements shift as content loads. A layout shift score above 0.1 means users are clicking on things that moved. Common causes: images without declared dimensions, dynamically injected content above the fold, web fonts loading after layout.

**Interaction to Next Paint (INP):** Responsiveness to user input — time between a user action (click, tap, keypress) and the next visual update. Replaced FID as the interaction metric. Target: under 200ms. Common causes of failure: long JavaScript tasks blocking the main thread.

**How much they matter:** Core Web Vitals are a ranking factor, but a marginal one. They are tiebreakers between comparable content, not primary determinants of rank. A fast page with weak content will not outrank a slow page with genuinely authoritative content. Where they matter most: highly competitive queries where many pages are comparable quality, and mobile results where speed differences are larger.

## Keyword cannibalization vs. topic clusters

These two concepts are opposite sides of the same problem:

**Cannibalization (bad):** Multiple pages competing for the same keyword phrase, splitting authority and confusing Google about which to rank. Symptoms: ranking fluctuations, both pages ranking but never one strongly, Google indexing the wrong page. Fix: consolidate into one definitive page, or differentiate intent clearly.

**Topic cluster (good):** Multiple pages covering a topic domain from different angles, with clear hierarchy (pillar → subtopic) and deliberate internal linking. Each page has a distinct primary intent; they support each other rather than compete. Symptoms of a working cluster: pillar ranks for head terms, cluster pages rank for long-tail variants, the pillar's authority lifts cluster pages.

**Diagnostic question:** If you put two pages side by side and can't clearly state how they serve different searcher intents, you have cannibalization. If you can state clearly why someone would want one vs. the other, you have a cluster.
<!--fold:2da7ad@file path="sources.md" mode="644"-->
# sources

Fetch these at task time. Ordered by relevance.

## Google's own documentation

1. Google Search Central — how search works, what Google's systems look for, official guidance:
   https://developers.google.com/search/docs

2. Google Search Central — Core Web Vitals and page experience signals:
   https://developers.google.com/search/docs/appearance/core-web-vitals

3. Google Search Central — Helpful Content system documentation:
   https://developers.google.com/search/docs/appearance/helpful-content-system

4. Google Quality Rater Guidelines (PDF) — the E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) that human raters use to evaluate pages:
   https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf

5. Google Search Central — understanding structured data and how it enhances search appearance:
   https://developers.google.com/search/docs/appearance/structured-data/intro-structured-data

## Methodology and frameworks

6. Ahrefs — Beginner's Guide to SEO, covering keyword research, on-page SEO, link building, and technical SEO:
   https://ahrefs.com/seo

7. Ahrefs — keyword research methodology, search intent classification, keyword difficulty:
   https://ahrefs.com/blog/keyword-research/

8. Moz — Beginner's Guide to SEO, comprehensive methodology reference:
   https://moz.com/beginners-guide-to-seo

9. Moz — Domain Authority and link quality metrics methodology:
   https://moz.com/learn/seo/domain-authority

10. Moz — on-page SEO factors, title tags, meta descriptions, heading structure:
    https://moz.com/learn/seo/on-page-factors

## Current algorithm updates and signals

11. Search Engine Journal — Google algorithm update coverage, confirmed updates and their impact:
    https://www.searchenginejournal.com/category/news/google-algorithm-updates/

12. Search Engine Journal — technical SEO guides and current best practices:
    https://www.searchenginejournal.com/category/seo/technical-seo/

13. Search Engine Roundtable — Barry Schwartz's real-time Google update tracking:
    https://www.seroundtable.com/

## John Mueller and Google communications

14. John Mueller's Search Central office-hours recordings (YouTube) — direct Q&A on how Google's systems work, common misconceptions:
    https://www.youtube.com/c/GoogleSearchCentral

15. Google Search Central Blog — official announcements of algorithm updates and new features:
    https://developers.google.com/search/blog

## Technical SEO

16. Web.dev — Core Web Vitals measurement and optimization guides:
    https://web.dev/explore/metrics

17. Screaming Frog SEO Spider documentation — crawl analysis methodology, identifying technical issues:
    https://www.screamingfrog.co.uk/seo-spider/user-guide/

## Link building and authority

18. Ahrefs — link building methodology, identifying link opportunities, evaluating link quality:
    https://ahrefs.com/blog/link-building/

19. Moz — link building guide, what makes a link valuable:
    https://moz.com/learn/seo/link-building

## AI in search

20. Google Search Central — AI Overviews, how Google's generative search experience works and how to appear in it:
    https://developers.google.com/search/docs/appearance/ai-overviews

21. Google Search Central — E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) and what "experience" signals mean for content quality:
    https://developers.google.com/search/docs/fundamentals/creating-helpful-content

22. Search Engine Journal — ongoing SGE/AI Overviews impact coverage on click-through rates and organic traffic:
    https://www.searchenginejournal.com/category/seo/ai-search/

23. Ahrefs — AI content and SEO: what the data shows about AI-generated content and rankings:
    https://ahrefs.com/blog/ai-content-seo/
<!--fold:2da7ad@end-->
PORTDOWN_24A5A142

# ── post ──
MARKER=$(awk '/^---$/ { f++; if (f==2) exit; next } f==1 && /^marker:[[:space:]]/ { sub(/^marker:[[:space:]]+/, ""); print; exit }' "$DEST")
[ -z "$MARKER" ] && { echo "seed: archive has no marker — corrupt" >&2; exit 1; }
awk -v m="$MARKER" -v outdir="$TARGET" '
  BEGIN {
    # Match <!--fold:<m>@file path="X"--> with an optional mode attr after
    # the path (fold emits  mode="644"  on executables).
    file_re = "^<!--fold:" m "@file path=\"([^\"]+)\"( mode=\"[0-9]+\")?-->$"
    end_re  = "^<!--fold:" m "@end-->$"
  }
  $0 ~ end_re { if (current) close(current); exit }
  $0 ~ file_re {
    if (current) close(current)
    line = $0
    sub(/^<!--fold:[^@]+@file path="/, "", line); sub(/".*$/, "", line)
    current = outdir "/" line
    dir = current; sub(/\/[^\/]*$/, "", dir)
    if (dir != current) system("mkdir -p \"" dir "\"")
    printf "" > current
    next
  }
  current { print >> current }
' "$DEST"
SEED_EXTRACTED=$(find "$TARGET" -type f -not -path "$DEST" 2>/dev/null | wc -l)
if [ "$SEED_EXTRACTED" = "0" ]; then
  echo "seed: archive contained no files — refusing to delete the source" >&2
  echo "  archive preserved at: $DEST" >&2
  exit 1
fi
rm -f "$DEST"

echo "" >&2
echo "✓ seed unpacked → $TARGET ($SEED_EXTRACTED files)" >&2
find "$TARGET" -type f | sort | while IFS= read -r _sf; do
  echo "  ${_sf#${TARGET}/}" >&2
done
echo "" >&2
if [ -f "$TARGET/SKILL.md" ]; then
  echo "This seed contains a skill (SKILL.md). Install it in your agent's skills directory." >&2
  echo "" >&2
fi
echo "Install the seed skill if not already installed:" >&2
echo "  https://seed.show/skill" >&2
echo "" >&2
echo "Publisher prompt:" >&2
sed 's/^/  /' >&2 <<'__SEED_PROMPT_END_AC1F2B__'
You have SEO strategy context across three files. Start with README.md: it covers what SEO is actually doing, the three-pillar framework, search intent taxonomy, what agents get wrong (including exact-match anchor text, PageRank proxies, and AI content failure modes), and how AI Overviews and E-E-A-T are changing the landscape. Read mental-models.md for the structural frameworks: intent taxonomy with diagnostic examples, the content→authority→trust flywheel, crawl budget management, site architecture as information architecture, Core Web Vitals as UX proxies, and keyword cannibalization vs. topic clusters. Fetch sources.md for authoritative current documentation — Google's own guidelines, methodology references, and AI-in-search coverage. Important: ranking factor weights are not public; any source claiming exact signal weights is extrapolating. To share your own context bundles: curl seed.show/skill | bash
__SEED_PROMPT_END_AC1F2B__
exit 0