Skip to main content

LiveWebsite is live

SEO and AI Search Analytics

SEO GEO: URL-Level AI Citation Analysis

Productionv0.1.0New

by Arsalan Usmani·Posted Sep 13, 2026

Diagnose why a single URL is or is not cited in AI Overviews and get page-level fixes that improve LLM citability.

Use this skillTalk to an expert

No credit card required

For one target URL, this skill pulls AI Overview citation data scoped to the URL's primary keywords from SE Ranking, identifies which AIO queries cite the page and which should but do not, and audits the page at passage level for citability. It compares your content against the passages competitors get cited for, checks schema and AI-protocol files when Firecrawl is available, and delivers a synthesised GEO.md report with concrete page-level recommendations. Distinct from domain-level share-of-voice analysis: this is one URL, deeper.

Capabilities

What it does

8 actions, each one checked against your rules and recorded before it runs.

validate_target

Validate the target URL and run preflight checks on connector availability and credit balance.

pull_keyword_footprint

Pull the URL's keyword footprint from SE Ranking and select the top traffic-weighted keywords.

detect_aio_presence

Query AI Overview presence and citation status for each keyword in the investigation set.

analyse_leaderboard

Retrieve the full AIO leaderboard per keyword to identify consistently cited domains and passage patterns.

audit_passages

Fetch the page and score each extractable passage for citability.

compare_cited_sources

Compare candidate passages against cited competitor passages to surface structural, content, and freshness gaps.

check_schema_protocols

Check JSON-LD schema and AI-protocol files such as llms.txt when Firecrawl is available.

synthesise_report

Synthesise findings and recommendations into the GEO.md report.

Step by step

How it runs

Every run follows the same governed path, from request to recorded result.

  1. 1

    Validate the target URL and confirm it is fetchable.

  2. 2

    Pull the URL's keyword footprint from SE Ranking and select the top 5 traffic-weighted keywords.

  3. 3

    Query AI Overview presence and citation lists for each keyword.

  4. 4

    Fetch the AIO leaderboard per keyword to see which domains are consistently cited.

  5. 5

    Fetch the page and score each extractable passage for citability.

  6. 6

    Compare uncited queries against cited competitor passages to surface structural gaps.

  7. 7

    Write GEO.md with findings and page-level recommendations.

Oversight

Governance & audit

What gets captured and verified on every single run.

Connector scope

Only SE Ranking and Firecrawl connectors are invoked. No access to user files, CRMs, or analytics properties beyond the target URL.

Firecrawl gating

Firecrawl usage is optional and gated. If unavailable or disabled, schema and AI-protocol steps emit explicit skipped notes rather than failing or inferring from markdown.

Credit budget

SE Ranking API usage is bounded to the URL keyword footprint plus AIO presence and leaderboard for the top 5 keywords, keeping per-run credit spend predictable.

No PII handling

Analysis is limited to publicly fetchable pages and public search data. No personal data is collected, stored, or included in reports.

Evidence retention

Run logs and evidence files are retained for 90 days, then purged. Reports contain no credentials or session data.

In practice

Example in action

SaaS blog post invisible in AI Overviews

  1. 1

    Preflight and footprint. The skill validates the URL, confirms it is fetchable, and pulls its top-5 traffic-weighted keywords from SE Ranking.

  2. 2

    Citation matrix. For each keyword, it checks AI Overview presence and whether the page is cited, then pulls the leaderboard to see which domains win citations.

  3. 3

    Passage audit. It fetches the page, extracts standalone passages, and scores each for citability: completeness, question fit, stats, and freshness.

  4. 4

    Gap comparison. For uncited queries, it fetches 2-3 cited competitor pages and compares passage shape against the candidate's content.

  5. 5

    Report delivery. It delivers GEO.md with the citation matrix, specific passage-level gaps, schema findings, and prioritised page-level fixes.

Under the hood

Skill manifest

The exact instructions this skill ships with, shown in full. Nothing hidden.

manifest.md
---
id: seo-geo
version: 0.1.0
display_name: SEO GEO: URL-Level AI Citation Analysis
pack: SEO and AI Search Analytics
category: marketing
risk_tier: low
tags: ['seo', 'geo', 'ai-overview', 'llm-citations', 'ai-search', 'se-ranking']

primary_domain: marketing
secondary_domains: ['data-analytics']
capabilities: ['web-research', 'search-knowledge-retrieval', 'reasoning-analysis', 'visualisation-reporting']
verticals: ['saas', 'media-publishing', 'retail-fmcg']
skill_type: composite
mechanisms: ['connector-based', 'cognitive', 'composite']
standards: []

input_schema:
  type: object
  required: [target_url]
  properties:
    target_url:
      type: string
      description: The full URL of the page to analyse for AI Overview citability.
    country:
      type: string
      description: Two-letter country code for search data scoping. Defaults to us.
    keywords:
      type: array
      description: Optional list of keywords to investigate. Defaults to the URL's top-5 traffic-weighted keywords from SE Ranking.
    no_firecrawl:
      type: boolean
      description: When true, skips Firecrawl-dependent steps even if Firecrawl is available, saving credits.

output_schema:
  type: object
  required: [geo_report, cited_keywords, uncited_keywords, recommendations]
  properties:
    geo_report:
      type: string
      description: Markdown content of the synthesised GEO.md report with findings and recommendations.
    cited_keywords:
      type: array
      description: Keywords where an AI Overview is present and the target URL is cited.
    uncited_keywords:
      type: array
      description: Keywords where an AI Overview is present but the target URL is not cited.
    recommendations:
      type: array
      description: Page-level changes recommended to improve LLM citability.
    evidence_files:
      type: array
      description: Paths to supporting evidence files such as passage audits and schema checks.

planner:
  mode: dynamic
  prompt_section: "How to think about this task"
  max_steps: 10
  max_duration_sec: 180
  max_cost_usd: 0.1
  model_tier: fast
  allowed_step_kinds: [tool_call, llm_call, output]

evidence:
  required_types: ['log']
  capture_per_step: true
  retention_days: 90

routing:
  intent_examples:
    - "Run a GEO analysis on this page."
    - "Why isn't this URL cited in AI Overviews?"
    - "Check AI search readiness for this URL."
    - "How do I improve LLM citations for this page?"
  not_for:
    - "Compare my brand's AI search share of voice against competitors at domain level."
    - "Run a full technical SEO site audit."

verifier_rules:
  - The report must state, for each investigated keyword, whether an AI Overview is present and whether the target URL is cited.
  - Every recommendation must reference a specific passage, structural element, or missing signal on the page.
  - If Firecrawl was unavailable, schema and AI-protocol sections must contain explicit skipped notes rather than inferred findings.
---

## Overview
Analyses a single URL's AI Overview citation footprint against its top keywords and recommends page-level changes to improve LLM citability. Pulls citation data from SE Ranking, audits the page at passage level, and compares against cited competitor passages. This is page-level diagnosis, not domain-level share of voice.

## How to think about this task
Start from the keywords, not the page. The unit of analysis is the AIO query: for each keyword the URL ranks for, an AI Overview either exists or does not, and the URL is either cited or not. That matrix drives everything downstream.

Work keyword by keyword. Pull AIO presence and the citation list, then the leaderboard to see which domains are consistently cited and what their cited passages look like. Patterns at the passage level matter more than domain-level authority.

Then audit the candidate page itself. Break it into extractable passages and score each for citability: complete thought in 1-3 sentences, answers the question its parent heading implies, contains a stat or named entity, carries a freshness signal. Compare that shape against the passages competitors get cited for.

Be honest about tooling limits. WebFetch returns markdown, so JSON-LD schema checks require Firecrawl. If Firecrawl is unavailable, record an explicit skipped note. Never infer schema from markdown.

## Capabilities
- Validate the target URL and run preflight checks on connector availability and credit balance.
- Pull the URL's keyword footprint from SE Ranking and select the investigation set.
- Detect AI Overview presence and citation status per keyword.
- Analyse the AIO leaderboard to identify consistently cited domains and passage patterns.
- Audit page passages and score each for citability.
- Compare candidate passages against cited competitor passages to surface gaps.
- Check JSON-LD schema and AI-protocol files when Firecrawl is available.
- Synthesise findings into a GEO.md report with page-level recommendations.

## Inputs
`target_url` is the page under diagnosis. It must be publicly fetchable; the run aborts early if it is not.

`country` scopes the search data to a market. It defaults to `us` and accepts a two-letter country code.

`keywords` is an optional override list. When omitted, the skill derives the investigation set from the URL's top-5 traffic-weighted keywords in SE Ranking.

`no_firecrawl` controls optional Firecrawl usage. When true, schema and AI-protocol steps are skipped even if Firecrawl is available, which saves credits.

## Outputs
`geo_report` is the primary deliverable: a synthesised Markdown report covering the citation matrix, passage audit, competitive gaps, schema and AI-protocol findings, and prioritised recommendations.

`cited_keywords` lists the keywords where an AI Overview exists and the target URL is already cited, so wins are preserved.

`uncited_keywords` lists the keywords where an AI Overview exists but the URL is absent. These are the optimisation targets.

`recommendations` is the ordered list of page-level changes, each tied to a specific passage, structural element, or missing signal.

`evidence_files` points to supporting artifacts such as the passage-level audit and the schema check, which editors consult when implementing changes.

## Tools
- connectors: SE Ranking Secure Connector (DATA_getUrlOverviewWorldwide, DATA_getDomainKeywords, DATA_getAiOverview, DATA_getAiOverviewLeaderboard), Firecrawl Secure Connector (firecrawl_scrape)
- operators: WebFetch
- primitives: passage extraction, citability scoring, JSON-LD parsing, markdown report synthesis

## Guardrails
- Only SE Ranking and Firecrawl connectors are invoked; no access to user files, CRMs, or analytics properties beyond the target URL.
- Firecrawl usage is optional and gated; when unavailable or disabled, schema and AI-protocol steps emit explicit skipped notes rather than failing or inferring from markdown.
- SE Ranking API usage is bounded to the URL keyword footprint plus AIO presence and leaderboard for the top 5 keywords, keeping per-run credit spend predictable.
- Analysis is limited to publicly fetchable pages and public search data; no personal data is collected, stored, or included in reports.
- Run logs and evidence files are retained for 90 days, then purged; reports contain no credentials or session data.

## Examples
A content lead at a SaaS company asks why their pricing-guide page never appears in AI Overviews. The skill validates the URL, pulls its top-5 keywords from SE Ranking, and finds AIOs present for four of them with the page cited for only one. The passage audit shows the page's answers are buried mid-paragraph with no standalone summaries after H2s, while cited competitors lead with 1-3 sentence definitional passages containing stats. The schema check confirms valid Article markup, and llms.txt returns 404. The GEO.md report recommends adding TL;DR summary blocks after each H2, inserting dated statistics into the two highest-traffic sections, and publishing an llms.txt file.
Feedback

Reviews

0.0(No ratings yet)

Reviews from teams using SEO GEO: URL-Level AI Citation Analysis.

No reviews yet

Be the first to share how SEO GEO: URL-Level AI Citation Analysis works for your team.

Discussion(0)

Sign in to join the discussion.Sign in
Failed to fetch

No comments yet. Be the first.