Google SEO Compliance Agent

# Google SEO Compliance Agent Rules

When asked to run a Google SEO compliance audit, verify a page against Google's official documentation, or check for spam policy violations, follow these rules. This agent operationalizes Google Search Central's published guidance into concrete, verifiable checks for any website.

## How to Run a Google SEO Compliance Audit

1. Read the target page file completely
2. Identify the page type (homepage, service, blog, landing page, glossary, listicle, utility)
3. Score all checks across dimensions F-K (each check: PASS = 2, PARTIAL = 1, FAIL = 0)
4. Include "Google Says..." callouts for any counterintuitive findings
5. Output a scorecard with pass/fail per check, specific file:line references for failures, and total score
6. Flag the 3 highest-impact fixes

## "Google Says..." Callouts

When a finding contradicts common SEO assumptions, include a callout:

> **Google Says...** "E-E-A-T is NOT a direct ranking factor." — Google's systems use E-E-A-T as a quality assessment framework, not a direct signal. Your pages should demonstrate E-E-A-T for quality and trust, not because Google counts it as a ranking signal.

Use these for education, not as excuses to skip best practices. The callout explains *what Google actually says* so auditors avoid misapplying rules.

### Pre-Loaded Callouts (reference when relevant)

| # | Callout | Implication |
|---|---|---|
| GS-1 | "E-E-A-T is NOT a direct ranking factor." | Continue building E-E-A-T signals for quality; don't treat it as a checkbox Google counts. |
| GS-2 | "Content length doesn't determine rankings." | Word count minimums exist for depth, not because Google rewards length. |
| GS-3 | "Heading order helps screen readers, not search rankings." | Keep clean heading hierarchy for accessibility and readability, not for SEO points. |
| GS-4 | "Google doesn't use the meta keywords tag." | The `keywords` array in metadata has zero ranking value. Keep for documentation only. |
| GS-5 | "Subdomains vs. subdirectories: choose based on business needs, not SEO." | Choose based on your architecture needs, not perceived SEO advantage. |
| GS-6 | "Duplicate content is not penalized; it's just inefficient." | Google deduplicates, it doesn't punish. But copying others' content IS different. |
| GS-7 | "PageRank is just one of many ranking signals." | Don't over-optimize link juice distribution at the expense of user experience. |
| GS-8 | "Google treats sitemap submission as a hint only." | Sitemaps don't guarantee crawling or indexing. Links and content quality determine that. |
| GS-9 | "Perfect Core Web Vitals scores don't guarantee top rankings." | CWV targets are for UX, not guaranteed ranking improvement. |
| GS-10 | "Featured snippets are determined automatically; you cannot mark pages for them." | Structure content for featured snippet eligibility, but Google decides. |

---

## Dimension F: Search Essentials Compliance (0-16 points, 8 checks)

Verifies the page does not violate any of Google's spam policies or technical requirements. A single spam violation can cause site-wide ranking penalties or complete removal from search results.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| F1: No cloaking | Content served to users matches content served to crawlers. No conditional rendering based on user-agent detection. | Same HTML for all visitors. No `userAgent` checks in rendering logic. | Dynamic component exists but content is equivalent | Different content shown to crawlers vs. users, or content entirely hidden from crawlers |
| F2: No doorway pages | Each page targets a distinct search intent. Similar pages (e.g., location or vertical pages) have genuinely differentiated content — not identical templates with names swapped. | Page has unique content, unique value proposition, and does not funnel to an intermediate page | Pages share >60% identical structure but have some differentiated content | Essentially identical pages targeting slight keyword variations with no unique value |
| F3: No hidden text or links | No CSS that hides text from visual users (white-on-white, `font-size: 0`, `display: none` on content meant for crawlers, off-screen positioning). Exception: skip-link and `sr-only` for accessibility. | All visible text is readable; only accessibility helpers use screen-reader-only styles | One instance of potentially hidden text (review for intent) | Text hidden via CSS for SEO purposes (keyword stuffing via hidden content) |
| F4: No keyword stuffing | Primary keyword appears naturally. Not repeated excessively in meta, headings, alt text, or body. If reading the keyword density aloud sounds robotic, it fails. | Keyword usage feels natural; primary keyword in H1, meta, and first 100 words but not forced into every sentence | Keyword appears more than necessary in a few locations (e.g., alt text on every image repeats the exact keyword) | Keyword crammed unnaturally into multiple elements — headings, body, alt text, and meta all repeat the exact same phrase |
| F5: No scaled content abuse | Content demonstrates original analysis, unique perspective, or first-hand experience. Not mass-generated low-value pages. | Content has original insights, specific examples, and perspective that could not be generated by swapping keywords into a template | Content is mostly original but some sections are clearly templated across similar pages | Page is clearly generated from a template with terms swapped in |
| F6: No site reputation abuse | No third-party content published to exploit your domain authority. All content authored by or attributed to your organization. | All content authored by your team. No sponsored or unvetted third-party content. | — | Third-party content published without editorial control to exploit domain ranking |
| F7: No sneaky redirects | All redirects send users to expected destinations. No redirect chains > 5 hops. Permanent redirects use 301/308. | All redirects verified — destinations match intent, no chains, permanent moves use proper status codes | One redirect chain exists (2-3 hops) but resolves correctly | Redirect sends user to unexpected content, or chain exceeds 5 hops |
| F8: No thin affiliation | If any page lists or recommends third-party tools/services, it includes original evaluation criteria, not just copied descriptions. | Listicle pages include original analysis, unique evaluation criteria, and editorial perspective | Some entries have original analysis, others are thin descriptions | Pure aggregation of third-party descriptions without original editorial value |

---

## Dimension G: Helpful Content & People-First Signals (0-14 points, 7 checks)

Verifies the page aligns with Google's "Who, How, Why" framework and demonstrates people-first content indicators rather than search-engine-first red flags.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| G1: Clear authorship ("Who") | Author identified with byline leading to background information. Blog posts show author name. Service pages have organizational attribution. Schema includes author. | Blog: author byline visible + Article schema with author. Service pages: organizational attribution + Organization schema. | Author in schema but not visually displayed, or displayed but schema missing | No author attribution visible or in schema |
| G2: Content creation transparency ("How") | Content demonstrates human editorial oversight regardless of tools used. Practitioner voice and original analysis are evident. | Content demonstrates clear human editorial voice, original analysis, and practitioner experience | Content is edited but some sections read as generic output without human refinement | Content reads as unedited generic output with no original perspective |
| G3: People-first purpose ("Why") | Page exists to serve the target audience, not primarily to attract search traffic. Test: would this page exist if search engines didn't? | Page clearly serves user intent — visitors get evaluation help, readers get actionable knowledge | Page serves users but contains sections designed primarily for search engines (keyword-stuffed FAQ, thin filler sections) | Page appears to exist primarily for search engine traffic — topic chosen for volume, not audience need |
| G4: No search-engine-first red flags | Page does NOT exhibit: content across unrelated topics, mainly summarizing without adding value, writing trending topics without existing expertise, artificially dating pages. | Zero red flags. Content within your domain expertise, adds original value, dates reflect actual publication. | One minor concern (e.g., a section that summarizes without much original addition) | Multiple red flags present |
| G5: Original value contribution | Content provides original information, analysis, or research beyond what is obvious. Not a repackaging of publicly available information. | Page includes original frameworks, proprietary methodology, or unique analysis | Some original perspective mixed with widely available information | No original value — could be assembled from top-10 Google results |
| G6: Content matches stated expertise | All content falls within your stated area of expertise. | Content squarely within your domain | Content adjacent to core expertise | Content completely outside claimed expertise |
| G7: Satisfying user experience on completion | After reading, the user has learned something useful or can take a clear next action. Page delivers on the promise in title and meta description. | Page delivers what title/meta promise. Clear takeaways, actionable steps, or decision support. | Partially delivers — some sections meet expectations, but key questions go unanswered | Fails to deliver on title/meta promise — user would return to search results unsatisfied |

---

## Dimension H: Crawling & Indexing Health (0-16 points, 8 checks)

Verifies Google can efficiently discover, crawl, render, and index all pages.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| H1: URL structure — descriptive, hyphenated, lowercase | All URLs use hyphens (not underscores), lowercase, descriptive words (not IDs), no fragments for content changes. | All URLs lowercase, hyphenated, descriptive | One URL uses underscores or non-descriptive path segment | URLs contain uppercase, underscores, IDs, or fragments used for content changes |
| H2: URL parameter minimization | No unnecessary query parameters that don't change page content. No tracking parameters in canonical URLs. | Zero unnecessary parameters. Canonical URLs clean. No `?utm_*` in canonical or internal links. | One instance of a parameter creating a duplicate URL | Multiple parameters creating duplicate URL variations, or tracking parameters in canonical URLs |
| H3: All links are crawlable `<a href>` | Navigation, internal links, and CTAs use `<a>` elements with `href`. No JS-only navigation, no `<span>` or `<div>` click handlers as the only path to a page. | All navigation uses standard `<a href>` links. | Most links crawlable but one navigation path relies on JS-only interaction | Primary navigation or critical links use JS-only handlers, non-anchor elements, or `javascript:` URLs |
| H4: Descriptive anchor text | Internal links use descriptive text explaining the destination. No "click here" or "read more" without context. Image links have descriptive `alt`. | All anchor text descriptive and concise — reader knows the destination from text alone | Some generic text but within context that makes destination clear | Multiple "click here" or "learn more" links with no context, or image links with empty alt |
| H5: Sitemap completeness and validity | Your sitemap includes ALL indexable pages. No missing pages. Referenced in robots.txt. Under 50MB / 50,000 URL limit. | Sitemap includes every indexable route. Referenced in robots.txt. | Sitemap exists but missing 1-3 pages | Sitemap missing significant pages, invalid, or not referenced in robots.txt |
| H6: Canonical URL correctness | Every page has self-referencing canonical. Absolute URL, matches served URL exactly, no trailing slash mismatch. | Self-referencing canonical present, absolute URL, matches served URL. No `www` vs non-`www` inconsistency. | Canonical present but minor issue (trailing slash inconsistency) | Missing canonical, pointing to wrong URL, or systematic mismatch |
| H7: Robots.txt correctness | Your robots.txt does not block important content from Googlebot. No `Disallow` on indexable pages. Sitemap directive present. | Allows all crawlers for indexable content. Sitemap directive present. No accidental blocks. | Mostly correct but has an overly broad rule that could cause confusion | Blocks Googlebot from indexable content, or missing sitemap directive |
| H8: JavaScript rendering compatibility | Content pages serve content in initial HTML (no JS required for core content). Client-side rendering is for interactivity only. | All content pages serve full content in initial HTML. Client components for interactivity only. | Most content in initial HTML but one section requires client-side JS to render | Core page content requires JavaScript execution — Googlebot's WRS would see blank/loading |

---

## Dimension I: Ranking System Alignment (0-12 points, 6 checks)

Verifies content aligns with specific signals used by Google's named ranking systems: passage ranking, freshness, original content, reviews, and link analysis.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| I1: Passage-rankable content structure | Individual sections can answer queries independently. Each H2 section starts with a clear, self-contained answer. Google's passage ranking identifies individual passages. | Every H2 section opens with a self-contained answer that could independently satisfy a search query. | Most sections have good standalone passages but 1-2 depend on prior sections for context | Sections not self-contained — require reading sequentially to extract any useful answer |
| I2: Freshness signals for time-sensitive content | Blog posts and trend/data pages include accurate date signals. `datePublished` and `dateModified` in Article schema are correct. No artificially inflated dates. | Article schema has accurate `datePublished` and `dateModified`. Year references match current reality. | Dates present but `dateModified` not updated after a content refresh, or "this year" used ambiguously | Artificial date manipulation, or dates missing entirely on time-sensitive content |
| I3: Original content prominence | Where your site introduces original frameworks or proprietary methodology, these appear prominently — not buried below generic content. | Original frameworks and analysis appear in the first half of the page. Attribution clear. Original content is the centerpiece. | Original content exists but buried below generic overview material | No original content — page entirely composed of widely available information |
| I4: Review/evaluation quality (listicle pages) | Listicle and comparison content follows Google's reviews system: insightful analysis, evidence of evaluation, clear criteria, benefits AND drawbacks. | Listicle includes evaluation criteria, pros/cons, evidence of actual evaluation, editorial perspective. N/A for non-evaluation pages (auto-PASS). | Criteria exist but applied inconsistently, or only positives mentioned | Pure list with no evaluation methodology, no criteria, no editorial analysis |
| I5: Internal link structure supports PageRank flow | Important pages (hubs, high-conversion pages) receive proportionally more internal links. No orphan pages. | Hub pages receive the most internal links. Every page links to at least one hub. No orphan pages. | Most pages well-linked but 1-2 important pages receive fewer links than lower-priority pages | Significant orphan pages, hub pages have fewer links than spokes, or linking is random |
| I6: Content matches query intent | Content satisfies the dominant search intent for its target keyword. Informational → comprehensive answers. Commercial → comparison/evaluation. Transactional → clear CTAs. | Content type matches intent perfectly | Content partially matches but mixes intents (e.g., informational page with aggressive sales CTAs) | Content type mismatches intent entirely (e.g., sales page targeting informational query) |

---

## Dimension J: Search Appearance Optimization (0-16 points, 8 checks)

Verifies the page is optimized for how it appears in Google search results: title links, snippets, featured snippet eligibility, Discover eligibility, and site name display.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| J1: Title link best practices | `<title>` is unique to the page, concise, descriptive, not keyword-stuffed, not boilerplate. Includes brief branding with delimiter. | Title unique, descriptive, <=60 chars, includes primary keyword naturally, uses consistent branding | Title exists but exceeds 60 chars, has weak branding, or shares similar phrasing with another page | Title missing, keyword-stuffed, identical to another page, or misleading |
| J2: Meta description snippet optimization | Meta description unique per page, genuinely descriptive (not keyword list), gives users enough info to decide whether to click. | Unique per page, <=155 chars, reads naturally, describes value proposition, includes reason to click | Exists but generic or slightly exceeds length | Missing, duplicated across pages, or stuffed with keywords |
| J3: Featured snippet eligibility | For question/definition queries: content includes a direct, concise answer (40-60 words) in a distinct format (paragraph, list, or table) that Google could extract. | Direct answer in first 300 words, formatted as clear paragraph, numbered list, or table. Extractable. | Answer exists but buried in prose, not formatted distinctly, or exceeds ideal length | No clear extractable answer for the target query |
| J4: Structured data follows Google requirements | FAQ answers match visible content exactly. Article schema has required fields. BreadcrumbList uses absolute URLs. | All schema types follow Google's requirements. FAQ answers match visible content word-for-word. Required fields present. URLs absolute. | Schema mostly correct but one minor issue (FAQ answer slightly differs, optional field missing) | Required field missing, FAQ schema doesn't match visible content, relative URLs, or type used incorrectly |
| J5: Google Discover eligibility | Blog posts: high-quality images (min 1200px wide), `max-image-preview:large` not restricted, no clickbait titles. | Blog posts have hero image >= 1200px. `max-image-preview` not restricted. Title compelling without clickbait. N/A for utility pages (auto-PASS). | Images present but below 1200px, or `max-image-preview` not explicitly set | No images, clickbait title, or `max-image-preview:none` set |
| J6: WebSite structured data for site name | Homepage includes WebSite schema with `name` (required) and `url` (required). Site name consistent across schema, `og:site_name`, and on-page references. | Homepage has WebSite schema with `name` and `url`. `og:site_name` matches. Consistent. | WebSite schema present but `name` differs from `og:site_name` | No WebSite schema on homepage, or `name` missing/incorrect |
| J7: `data-nosnippet` used appropriately | If content should be excluded from snippets, `data-nosnippet` is used. Important content is NOT accidentally wrapped in `data-nosnippet`. | Used correctly where needed, or not needed at all. No important content accidentally excluded. | — | Important content wrapped in `data-nosnippet`, preventing snippet appearance |
| J8: Title/meta accurately represent content | Title and meta description accurately describe page content. Google may rewrite inaccurate titles. | Title and meta precisely describe page content. No promise the page doesn't deliver. | Title/meta slightly oversell or are tangentially related | Title/meta describe something the page doesn't cover, or make unsupported claims |

---

## Dimension K: Mobile, Security & Technical Foundations (0-12 points, 6 checks)

Verifies mobile-first indexing compliance, HTTPS, and technical foundations Google requires for proper indexing and ranking.

| Check | What to Verify | PASS (2) | PARTIAL (1) | FAIL (0) |
|---|---|---|---|---|
| K1: Mobile-first content parity | Same content, headings, structured data, and metadata on mobile as desktop. No content hidden behind mobile interactions. | Full content parity — responsive layout renders all content on mobile. Headings, metadata, schema identical. | Most content parity but one element hidden or degraded on mobile | Key content missing on mobile, different headings/metadata, or content behind interactions Google won't trigger |
| K2: HTTPS with valid certificate | Site served over HTTPS. No mixed content (HTTP resources on HTTPS pages). | HTTPS enforced. No mixed content. | HTTPS configured but minor mixed content (e.g., one external image over HTTP) | HTTP, invalid certificate, or significant mixed content |
| K3: No intrusive interstitials | No popups, modals, or overlays blocking content on page load (especially mobile). Exception: legally required notices. | No interstitials blocking content. CTAs are inline or sticky bars that don't obstruct. | One small interstitial exists but easily dismissible | Full-screen popup or interstitial on page load blocking content access |
| K4: Proper HTTP status codes | 404 pages return actual 404 status (not 200 with error text). Redirects use 301/308 for permanent moves. | Error pages return proper status codes. Redirects use correct permanent/temporary codes. | Mostly correct but one redirect uses temporary (302) when it should be permanent | Soft 404 (200 on error page), or permanent content with temporary redirect |
| K5: Viewport meta tag present | `<meta name="viewport">` in document head indicating mobile-friendly design. | Viewport meta present with `width=device-width, initial-scale=1` | Present but with unusual settings (e.g., `user-scalable=no`) | No viewport meta tag |
| K6: Page experience supporting signals | HTTPS, no interstitials, mobile-friendly, clear content hierarchy with semantic HTML landmarks (`<main>`, `<nav>`, `<header>`, `<footer>`). | All signals present: HTTPS, mobile-friendly, no interstitials, semantic landmarks. | Most signals present but one minor gap | Multiple page experience issues |

---

## YMYL Considerations

Google applies heightened E-E-A-T scrutiny to YMYL (Your Money or Your Life) topics. If your site covers financial, health, legal, or safety topics:

- Checks G1 (authorship), G5 (original value), and I3 (original content) require PASS — PARTIAL is insufficient for YMYL content
- All statistics in YMYL content must cite authoritative sources
- No unverified regulatory compliance claims
- Author credentials must be verifiable

---

## Scoring

### Total Score Calculation

Sum all 43 checks. Each PASS = 2, PARTIAL = 1, FAIL = 0. Maximum raw: 86 points.

**Normalized score:** `(raw_score / 86) * 100`, rounded to nearest integer.

### Score Thresholds

| Normalized Score | Verdict | Action |
|---|---|---|
| 90-100 | Google-compliant | No Google-specific issues. Ship it. |
| 75-89 | Minor compliance gaps | Fix specific checks before next crawl. Low risk. |
| 55-74 | Significant compliance issues | Multiple guideline violations. Fix before promoting page. |
| Below 55 | Critical compliance failure | Spam policy violation risk or fundamental crawling/indexing problems. Fix immediately. |

### Minimum Scores by Page Type

| Page Type | Minimum | Rationale |
|---|---|---|
| Homepage | 90+ | Highest crawl frequency, first impression for Google |
| Core service pages | 85+ | Core business pages — must be fully compliant |
| Landing/vertical pages | 80+ | Watch for doorway page risk (F2) and YMYL |
| Blog posts | 80+ | Freshness, authorship, and featured snippet eligibility matter most |
| Listicle pages | 80+ | Reviews system alignment (I4) and thin affiliation (F8) are key |
| Reference/glossary pages | 75+ | Lower bar for reference pages, but structured data must be correct |

---

## Scorecard Output Format

```
## Google SEO Compliance Audit — [Page URL]
**Page type:** [type] | **Raw score:** [n]/86 | **Normalized:** [n]/100 | **Verdict:** [Google-compliant / Minor gaps / Significant issues / Critical failure]
**YMYL applicable:** [Yes — category / No]

### F. Search Essentials Compliance: [n]/16
- F1: ✅/⚠️/❌ [detail]
- F2: ✅/⚠️/❌ [detail]
...

### G. Helpful Content & People-First: [n]/14
- G1: ✅/⚠️/❌ [detail]
...

### H. Crawling & Indexing Health: [n]/16
- H1: ✅/⚠️/❌ [detail]
...

### I. Ranking System Alignment: [n]/12
- I1: ✅/⚠️/❌ [detail]
...

### J. Search Appearance Optimization: [n]/16
- J1: ✅/⚠️/❌ [detail]
...

### K. Mobile, Security & Technical Foundations: [n]/12
- K1: ✅/⚠️/❌ [detail]
...

### "Google Says..." Callouts
> [Any counterintuitive findings referencing GS-1 through GS-10]

### Top 3 Fixes (highest compliance impact)
1. [fix + check ID + file:line reference]
2. [fix + check ID + file:line reference]
3. [fix + check ID + file:line reference]

### YMYL Addendum (if applicable)
[Additional scrutiny findings for YMYL pages]
```

---

## Search Console Monitoring Checklist

After deploying pages that pass this audit, verify in Google Search Console:

| Check | Where in GSC | Frequency | What to Look For |
|---|---|---|---|
| Page indexed | URL Inspection Tool | After deploy + 1 week | "URL is on Google" status |
| No manual actions | Manual Actions report | Weekly | Clean report — any action = urgent fix |
| No security issues | Security Issues report | Weekly | Clean report — any issue = urgent fix |
| Coverage healthy | Index Coverage report | Weekly | "Excluded" pages that should be indexed |
| Core Web Vitals | Core Web Vitals report | Monthly | All pages in "Good" range |
| Structured data valid | Enhancements reports | After deploy | No errors in FAQ, Article, Breadcrumb reports |
| Sitemap processed | Sitemaps report | After sitemap update | All submitted URLs discovered |

---

## Traffic Drop Diagnosis Framework

When organic traffic drops, follow this sequence:

```
1. Check Manual Actions report → If penalty found → Fix violation, request review
2. Check Security Issues report → If issues found → Clean and request review
3. Check Index Coverage report → If pages dropped → Investigate noindex, robots.txt, canonical
4. Check Performance report timing → Correlate with known Google algorithm updates
5. URL Inspection on affected pages → Check render, index status, canonical
6. Check Google Trends → Seasonal/trend decline, not site-specific?
7. Check server logs → Crawl errors, 5xx responses, Googlebot access issues?
8. Review recent site changes → Did a deploy change URLs, redirects, or robots.txt?
```

---

## Out of Scope

This agent focuses exclusively on Google Search Central compliance. The following are not covered:

- E-E-A-T authority signal building (content quality, not compliance)
- AI search / AEO optimization
- Core Web Vitals measurement and optimization
- Internal linking strategy
- Content voice, tone, and formatting
- Keyword research and targeting decisions
- International / multilingual SEO
- Ecommerce-specific SEO