AI GEO Recommendation Tracker — Audits & Reports
Pricing
from $40.00 / 1,000 prompt x engine checks
Audit AI visibility across ChatGPT, Claude, Gemini, and Perplexity. Get prioritized GEO recommendations, content gaps, trends, and HTML reports.
The brand, company, or product to track in AI answers (e.g., 'Notion', 'Tesla'). Used as the primary entity for mention detection, position ranking, share-of-voice computation, and recommendation generation.
Your brand's website domain (e.g., 'notion.so'). Used to detect when AI engines cite your site as a source and to compare citation rates against competitor domains. Leave empty to skip domain-based citation matching.
Your category (e.g., 'productivity software', 'electric vehicles'). Used to auto-generate a relevant prompt set when you do not supply your own prompts below. Also informs recommendation context for GEO improvement suggestions. Ignored if 'prompts' is provided.
The exact questions your target audience asks AI engines (e.g., 'What is the best note-taking app?'). Provide your own list for precise tracking. If left empty, prompts are auto-generated from 'industry'. Each prompt is checked on every selected platform. Higher prompt variety produces richer recommendations.
[]Which AI answer engines to query. All engines run through OpenRouter using a single OPENROUTER_API_KEY (set as an environment variable). Default models: perplexity→perplexity/sonar, chatgpt→openai/gpt-4o-mini, claude→anthropic/claude-sonnet-4, gemini→google/gemini-2.5-flash. Override slugs via the 'models' field.
[ "perplexity", "chatgpt", "claude", "gemini"]Competitor brand names to track alongside yours (e.g., 'Evernote', 'Obsidian', 'Coda'). Used to compute your position ranking, share-of-voice, and to identify which competitors are outperforming you in AI visibility. The more competitors listed, the richer the comparison insights and recommendations.
[]Competitor website domains for citation comparison (e.g., 'evernote.com', 'obsidian.md'). Used to detect when AI engines cite competitor sites but not yours — a key signal for GEO recommendations. Leave empty if you do not need domain-level citation gap analysis.
[]When enabled, produces an agency-ready HTML report in the key-value store under REPORT. The report includes visibility score cards, per-platform breakdowns, position trends (if history exists), citation analysis, and top prioritized recommendations. Disable for MCP/API-only consumption to save runtime.
Optional stable identifier used to store and compare history across scheduled runs. Defaults to a slug of the brand name. Use different IDs to track multiple brands independently. The history enables trend-based recommendations (e.g., 'visibility has dropped 3 weeks in a row — take action now').
Enable live web search for models that support it via OpenRouter's web plugin (Perplexity Sonar already searches natively). Keep on for accurate AI-search visibility based on current web data. Turn off to test the models' built-in knowledge only.
How many web results the OpenRouter web plugin should fetch per query. More results produce richer answers with more citations but increase cost per query. Applies to non-native-search models (ChatGPT, Claude, Gemini).
Optional. Map each platform to a specific OpenRouter model slug for custom model selection. Example: {"chatgpt": "openai/gpt-4o", "claude": "anthropic/claude-sonnet-4"}. Any platform not listed uses its default model. Use when you want to test visibility across different model versions.
{}Safety cap on how many prompts are checked per run. Controls cost and runtime. Extra prompts beyond this limit are silently ignored. Higher values produce more granular recommendations but increase API cost. Free users receive at most 1 result from 1 OpenRouter request per run.