AI Overviews Tracker — Brand Visibility in AI Answers
Pricing
$90.00 / 1,000 keyword × query × platform checks
Track brand mention rate, sentiment and share of voice in AI-generated answers from Perplexity, ChatGPT, Claude and Gemini. Built for GEO audits and brand monitoring through Apify MCP or the API.
Your brand, product, or company name to track in AI-generated overviews. The actor detects mentions, recommendation phrases, and sentiment for this exact name. Example: 'HubSpot', 'Notion', 'Figma', 'Stripe'. NOT a domain name (use ai-search-visibility-tracker for domain citation tracking) — this field expects the brand name humans and AI would use in sentences.
Purchase-intent or commercial keywords to submit to AI platforms as overview queries. Each keyword generates queries like 'best [keyword] tools' or 'how to choose [keyword]'. Example: ['best CRM for small business', 'email marketing software', 'project management tools']. NOT brand names — enter the product category or use case your brand serves. Supports up to 20 keywords per run.
Competitor brand names to compare against your brand in AI overviews. The actor tracks whether each competitor is mentioned, recommended, and at what position — and computes share of voice (your brand's mentions vs total brand+competitor mentions). Example: ['Salesforce', 'Pipedrive', 'Monday.com']. NOT domains — enter brand names as they appear in sentences. Leave empty to skip competitor comparison.
Which AI platforms to check for brand visibility in overviews. Valid values: 'perplexity' (Sonar with real-time citations), 'chatgpt' (OpenAI web search), 'claude' (Claude Sonnet with web_search tool), 'gemini' (Gemini 2.5 Flash with Google Search grounding). Defaults to Perplexity, ChatGPT, and Gemini. Add 'claude' explicitly for deep audits.
[ "perplexity", "chatgpt", "gemini"]Built-in purchase-intent query templates to generate for each keyword. Options: 'best_of' (best [keyword] solutions in 2026), 'comparison' (how do leading [keyword] platforms compare), 'how_to_choose' (how to select the right [keyword] for your business), 'buyer_intent' (what [keyword] do most companies use), 'feature_query' (what features matter in a good [keyword]), 'versus' (your brand vs alternatives for [keyword] — requires brandName). Defaults to best_of, comparison, how_to_choose.
[ "best_of", "comparison", "how_to_choose"]Specific questions to submit verbatim to all AI platforms for every keyword. Supports {keyword} and {brand} placeholders (e.g., 'What do industry analysts say about {brand} for {keyword}?'). These run in addition to queryTemplates. Example: 'What are the most recommended {keyword} tools for enterprise use?'. NOT URLs — provide full question strings in plain English.
Maximum number of distinct queries to run per keyword across all templates. Each query × platform = one check billed at $0.096. Default is 3 queries/keyword — covers best_of + comparison + how_to_choose at one query each. Set to 5–6 for deeper coverage including buyer_intent and feature_query. Lower to 1–2 for fast overview spot-checks.
Controls the length of the overview_excerpt field stored per result. 'concise' stores the first 300 characters of the AI response — faster to review and lower storage. 'detailed' stores the first 550 characters — better for reading the full brand context and competitor mentions. Does not affect what is tracked or scored — only the excerpt length.
An HTTPS endpoint that receives a POST request when the run completes. Payload is the full run summary JSON including brand_visibility_score, overview_presence_score, competitor_threats, share_of_voice map, keyword_scores, and recommendations. Compatible with Make.com, Zapier, n8n, Slack incoming webhooks, and any HTTP endpoint that accepts JSON.
Set to true to verify the actor is functional without making any AI API calls and without consuming credits. The actor exits immediately with a success message. Use for integration testing, scheduling health checks, or MCP tool validation in CI pipelines. Set to false (default) for real brand visibility tracking runs.