Programmatic SEO is the creation and management of search pages at scale from structured data and reusable templates. Claude SEO audits data sources, template quality, URL patterns, internal links, canonicals, sitemaps, index bloat, and safeguards against scaled content abuse.
Audit an existing programmatic surface or plan a staged batch from a lawful, reviewed data source.
OUTCOME
An evidence-backed programmatic audit or staged plan, never a guarantee that a batch is safe to publish.
REQUIRED INPUT
A site URL or the plan route in /seo programmatic <url or plan>, plus the relevant template and data context.
READINESS
Data must be lawful, current, and available through least-privilege read access with rate limits and provenance.
LOCAL OUTPUT
Audit findings or a rollout plan covering URL, canonical, linking, index, data-quality, and sitemap decisions.
SIDE EFFECTS
Can query files, APIs, or databases and produce local analysis. Downstream page generation and production publication are separate actions.
COST + APPROVAL
Confirm any provider cost first. Data use, batch size, canonical logic, generated content, and publication require human approval.
// HOW IT WORKS
What does a programmatic SEO audit cover?
When you run /seo programmatic, Claude SEO analyzes programmatic pages across data source quality, template value, URL structure, internal linking, thin-content risk, index management, canonical strategy, and sitemap integration. The scores and stop conditions are internal workflow controls, not Google thresholds or penalty predictions.
The tool helps reviewers apply Google's scaled content abuse policy, which focuses on generating many pages primarily to manipulate rankings rather than help users. Google does not publish numeric uniqueness, word-count, rollout, or sample-size thresholds. Claude SEO's conservative defaults are review prompts, and every page still needs a useful standalone purpose.
What each analysis area covers
The analysis areas produce a 0-100 internal prioritization score. Review gates can pause a large or weakly differentiated plan for explicit human approval. The score does not come from Google and does not predict indexing, rankings, manual actions, or traffic.
01
DATA SOURCE QUALITY
Evaluates CSV/JSON files, API endpoints, and database queries. Reviews row count, column uniqueness, missing values, duplicate-record indicators, freshness, and rate limits. Sources require explicit read permission and least-privilege credentials.
02
TEMPLATE UNIQUENESS
Reviews variable injection points, static vs dynamic content blocks, conditional logic, and supplementary content. No "mad-libs" patterns. Each page must read as a standalone resource.
03
URL PATTERNS
Reviews URL structure for readable slugs, logical hierarchy, duplicates, unnecessary length, and avoidable parameters. Supports /tools/, /city/service/, /integrations/, /glossary/, and /templates/ patterns.
04
INTERNAL LINKING
Reviews category hubs, contextual related-page links, breadcrumbs, and cross-links between pages sharing useful attributes. Link counts should follow reader needs, not a fixed quota.
05
THIN CONTENT RISK
Uses conservative vendor gates to surface large unreviewed sets, repeated templates, weak differentiation, and pages that lack standalone value. These are review triggers, not Google limits.
06
INDEX MANAGEMENT
Keeps useful paginated pages crawlable with distinct URLs and self-canonicals. Reserves noindex or canonical consolidation for true duplicates and low-value filtered views. Monitors indexed versus intended pages on large sites.
07
CANONICAL STRATEGY
Uses self-referencing canonicals for distinct pages. Duplicate or low-value parameter variations can canonicalize to a representative URL. Overlap with a manual page requires evidence-based consolidation, not an automatic rule.
08
SITEMAP INTEGRATION
Auto-generates sitemap entries. Split at 50,000 URLs per file. Sitemap index for multiple files. lastmod reflects actual data update timestamp. Excludes noindexed pages. Registered in robots.txt.
SPAM POLICY
SCALED CONTENT ABUSE
Checks whether pages are created primarily to manipulate rankings instead of helping users. Recommends staged publication and human review without presenting vendor defaults as Google policy.
// SCORING
How is programmatic SEO scored?
The programmatic SEO score is a Claude SEO heuristic across 6 categories. Each category shows a status indicator for prioritization. The weights are vendor choices, not search-engine signals.
DATA QUALITY
20%
TEMPLATE UNIQUENESS
25%
URL STRUCTURE
10%
INTERNAL LINKING
15%
THIN CONTENT RISK
20%
INDEX MANAGEMENT
10%
Candidates and high-risk patterns
Candidates for review at scale: integration pages with real setup documentation, useful template pages, accurate glossary entries, product pages with verified specifications, and data-driven pages with meaningful record-level evidence. No word count makes a page safe. Each page still needs accurate data, standalone value, and human review.
High review priority: location pages with only a city name swapped, "best [tool] for [industry]" pages without industry-specific value, "[competitor] alternative" pages without real comparison data, and large batches published without human review. Shared boilerplate can be legitimate, so assess whether each page adds useful, accurate, standalone value.
Claude SEO detects template-generated patterns, evaluates data quality, compares page differentiation, and applies transparent vendor review gates. You get an internal prioritization score with the evidence behind each recommendation.
Step 4: Or plan new programmatic pages
/seo programmatic plan
When planning mode is used, Claude SEO helps you design the data source, template engine, URL patterns, internal linking, and rollout strategy before generating any pages. Use only lawful, current data with explicit read permission, respect rate limits, and keep credentials least-privileged. Analysis and drafting do not authorize publication; sample and approve pages before launch.
Default vendor review gates
These defaults pause or flag a workflow for human review. They are not limits published by Google and should be adjusted to the project's evidence and risk.
Metric
Threshold
Action
Pages without content review
100+
WARNING: require content audit before publishing
Pages without justification
500+
HARD STOP: require explicit approval and thin content audit
Page differentiation estimate
<40%
Flag for standalone-value review
Short page heuristic
<300 words
Review purpose and completeness, not length alone
// FAQ
QUESTIONS ABOUT PROGRAMMATIC SEO
Claude SEO uses conservative vendor gates to flag large unreviewed page sets, repeated templates, weak differentiation, missing evidence, and pages without standalone value. These gates can pause a workflow for human review, but they are not Google thresholds or predictions of a penalty.
The tool evaluates CSV and JSON files, API endpoints, and database queries. It reviews structure, completeness, duplicates, freshness, and rate limits. Use only lawful, current sources with explicit read permission and least-privilege credentials; analysis does not authorize publishing.
No tool can guarantee that. Claude SEO helps reviewers check whether each page has a useful purpose, accurate source data, meaningful differentiation, and human oversight. Google's policy is about content created at scale primarily to manipulate rankings, and Google does not publish numeric uniqueness or word-count thresholds.
The tool can plan patterns such as /tools/[tool-name], /[city]/[service], /integrations/[platform], /glossary/[term], and /templates/[template-name]. It recommends readable, stable URLs and consistent hierarchy, while allowing justified exceptions for a platform's routing needs.
Claude SEO's code is open-source under the MIT license. Running the analysis still requires Claude Code or model access under the user's plan. The skill has no separate paid feature tier and does not publish pages without human approval.
// RELATED SKILLS
EXPLORE MORE
SITEMAP
XML sitemap analysis, quality gates, format validation, and HTTP status verification for all listed URLs.
