AI GEO Recommendation Tracker — Visibility Audit & Reports

Run a complete GEO audit across ChatGPT, Claude, Gemini, and Perplexity. The actor measures brand mentions, AI search position, share of voice, citations, sentiment, and trend changes, then produces prioritized AI visibility recommendations and an agency-ready HTML report.

Built for SEO agencies, GEO consultants, brand teams, and content strategists who need both measurement and a concrete improvement plan—not another dashboard with unexplained scores.

What It Does

Queries ChatGPT (GPT-4o-mini), Claude (Sonnet 4), Gemini (2.5 Flash), and Perplexity (Sonar) through a single OpenRouter API key with your tracked prompts. For each prompt × engine combination, it:

1. Detects brand mentions — is your brand named? How many times? In what context?
2. Ranks your position — among you and your listed competitors, where do you appear first?
3. Computes share-of-voice — what percentage of the tracked-brand mentions are yours?
4. Extracts citations — what sources does the AI cite? Is your domain among them?
5. Scores sentiment — positive, neutral, or negative classification of how you're described.
6. Compares vs previous runs — gained, lost, improved, or declined across tracked keywords and engines.
7. Generates GEO recommendations — a prioritized action plan with concrete steps: what content to create, which authority signals to build, and where the gaps are.
8. Produces a shareable HTML report — visibility score cards, per-platform breakdowns, citation analysis, and all recommendations in a self-contained file.

What makes this GEO audit different?

Most AI visibility trackers stop at mentions. This actor connects observed answer-engine data to a rule-based recommendation layer. It identifies visibility gaps, missing citations, weak competitive position, sentiment issues, and uncovered topics, then saves the evidence and recommended actions separately so agencies can audit the logic behind the report.

Use the dataset for detailed analysis, SUMMARY for scorecards, RECOMMENDATIONS for an action backlog, and REPORT for client delivery.

Input

Field	Type	Required	Description
brandName	string	Yes	Brand/company/product to track (e.g., "Notion")
brandDomain	string	No	Your website domain for citation detection (e.g., "notion.so")
industry	string	No	Category for auto-generating prompts (e.g., "productivity software")
prompts	string[]	No	Exact questions to monitor; auto-generated from industry if empty
competitorBrands	string[]	No	Competitor brand names for position and share-of-voice comparison
competitorDomains	string[]	No	Competitor domains for citation gap detection
generateHtmlReport	boolean	No	Produce an HTML report in key-value store (default: true)
trackingId	string	No	Stable ID for cross-run history comparison
maxPrompts	integer	No	Cap on prompts per run (1-200, default: 25)
useWebSearch	boolean	No	Enable live web search via OpenRouter plugin (default: true)
webSearchResults	integer	No	Number of web results per query (1-10, default: 3)
models	object	No	Per-platform OpenRouter model slug overrides

Output

Dataset (one row per prompt × engine)

Each record contains: brand, brand_domain, engine, model_used, prompt, mentioned, position, share_of_voice, cited, citations, brand_cited_url, mention_count, mention_context, sentiment, competitors_mentioned, topic_keywords, citation_domains, competitor_domains_cited, previous_mentioned, previous_position, position_change, visibility_status, error, checked_at

Key-Value Store

Key	Type	Description
SUMMARY	JSON	Visibility score, per-engine breakdown, trend deltas, top cited domains
RECOMMENDATIONS	JSON	Prioritized action plan with 5 categories, severity levels, findings, and impact estimates
REPORT	HTML	Agency-ready self-contained report (when generateHtmlReport is true)

Recommendation Categories

Category	What It Detects	Example Action
Visibility	Brand absent from checks	Create content targeting missing topics
Citation	Domain not cited, competitors are	Build authority with AI-trusted domains
Sentiment	Negative or neutral mentions	Improve brand perception and reviews
Position	Low rank among competitors	Strengthen E-E-A-T and authority signals
Coverage	Visible for some prompts but not others	Fill content gaps in uncovered topics

Pricing

This actor uses Pay-Per-Event pricing:

Event	Price	Description
actor-start	$0.00005	One-time initialization fee
prompt-engine-check	$0.025	Per prompt × engine check (all analysis, citations, and recommendations included)

Example: 4 prompts × 4 engines = 16 checks × $0.025 = $0.40 per run.

OpenRouter model and web-search costs are paid through your own OpenRouter account and are separate from Apify Pay-Per-Event charges. Control the run with maxPrompts, webSearchResults, and the prompts you provide. Check the live Store pricing panel before production use.

Environment Variables

Variable	Required	Description
OPENROUTER_API_KEY	Yes	Your OpenRouter API key for unified multi-LLM access

Schedule It

Set up a daily or weekly schedule in Apify Console to build a visibility history. The actor compares each run against the previous one, producing trend data (gained/lost/improved/declined) that surfaces negative movement before it compounds.

Use Cases

• Agencies — Produce client-ready GEO reports with concrete recommendations and visual scorecards
• SEO/GEO teams — Identify content gaps, authority weaknesses, and citation opportunities across AI engines
• Brand marketers — Monitor how AI answer engines describe your brand vs competitors
• MCP/AI agents — Clean structured output with semantic field names for programmatic consumption

Run the GEO audit through the API

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('USERNAME/ai-geo-rec-tracker').call({
  brandName: 'Notion',
  brandDomain: 'notion.so',
  industry: 'productivity software',
  prompts: ['What is the best workspace tool for startups?'],
  competitorBrands: ['Evernote', 'Obsidian', 'Coda'],
  competitorDomains: ['evernote.com', 'obsidian.md', 'coda.io'],
  trackingId: 'notion-geo-audit',
  generateHtmlReport: true,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

Use Apify schedules for weekly or monthly audits, webhooks for downstream delivery, dataset exports for analysis, or Apify MCP for agent-driven research.

When to use—and when not to use—this actor

Use it when the desired outcome is an evidence-backed GEO action plan, multi-client report, content-gap backlog, or recurring AI-search audit. Choose the simpler AI Visibility Rank Tracker when you only need raw monitoring metrics. Choose the AI Citation Monitor when the primary goal is a detailed inventory of cited domains and URLs.

This is not a conventional Google keyword rank tracker, website crawler, backlink index, or guarantee that recommendations will change model answers. The recommendation engine interprets the collected checks; it cannot observe every prompt or every personalized response.

Best practices for stronger recommendations

• Supply real buyer questions and category prompts, not only branded queries.
• Add direct competitors and their domains for useful position and citation-gap analysis.
• Keep the same trackingId and inputs across scheduled audits.
• Review error rows before treating missing visibility as a content problem.
• Prioritize repeated gaps across multiple prompts and engines over a single anomalous answer.

FAQ

What is a GEO audit?

A Generative Engine Optimization audit measures how AI answer engines mention, position, describe, and cite a brand, then translates those observations into actions that may improve AI discoverability.

Are the recommendations generated by another paid LLM call?

The recommendations are rule-based from the collected metrics and are included in the prompt-engine-check event described above.

Can an agency create reports for multiple clients?

Yes. Use a unique trackingId for every brand or client and keep the prompt set stable. The optional REPORT artifact is designed for sharing, while JSON outputs support custom dashboards.

Reliability and responsible use

AI answers change as models, search indexes, and source availability change. Treat scores and recommendations as decision support, validate important findings manually, and avoid promising clients guaranteed placement. Store OPENROUTER_API_KEY as a secret. Use the actor for lawful research and comply with applicable privacy, marketing, and platform requirements.

Documentation & Integration

Export scraped data, run the scraper via API, schedule and monitor runs, or integrate with other tools. Use the Apify API to fetch results programmatically from your dashboard, data pipeline, or LLM agent.

Get the OpenRouter API key to start tracking.