AI Overviews Tracker — Brand Visibility in AI Answer Boxes

Track whether your brand is mentioned, recommended, or excluded from AI-generated overview boxes across Perplexity, ChatGPT, Claude, and Gemini for your commercial keywords. Measure brand mention rate, recommendation rate, sentiment, and share of voice against competitors — with actionable GEO (Generative Engine Optimization) recommendations.

What It Does

When someone searches "best CRM software" or "top email marketing tools", AI platforms like Perplexity, ChatGPT, Claude, and Gemini generate rich overview boxes recommending specific products. This actor tells you:

• Is your brand in those overviews? Exact mention detection with sentence-level position tracking.
• Are you recommended or just mentioned? Detects recommendation language ("top pick", "highly recommended", "consider") vs passive mentions.
• What's the sentiment? Positive, neutral, or negative framing near your brand name.
• Who's outranking you? Competitor share of voice: your brand's mentions vs competitor mentions in the same overviews.
• Which keywords expose your gaps? Full per-keyword breakdown showing where you're invisible.

Who This Is For

• SaaS founders and product marketers tracking whether their product is recommended in AI overviews for their category keywords
• SEO teams doing GEO (Generative Engine Optimization) who need data on which commercial queries trigger AI overviews and whether their brand is included
• Paid search managers monitoring competitor presence in organic AI overviews that compete with their paid ads
• Brand managers tracking brand sentiment in AI-generated content at scale

Not the same as our other AI tracking actors:

• AI Brand Monitor — tracks brand name mentions in informational AI answers (reputation/PR focus)
• AI Search Visibility Tracker — tracks whether your domain is cited as a source (SEO/AEO focus)
• AI Overviews Tracker (this actor) — tracks brand presence in commercial intent AI overview boxes (GEO/marketing focus)

Output Data

Each record represents one keyword × query × platform check:

Field	Type	Description
platform	string	AI platform: perplexity, chatgpt, claude, gemini
keyword	string	Commercial keyword submitted
query	string	Full query text sent to the AI
query_template	string	Template type: best_of, comparison, how_to_choose, etc.
overview_triggered	boolean	Did the AI generate a rich overview box?
overview_type	string	comparison, recommendation, explanation, or none
brand_mentioned	boolean	Was your brand found in the overview?
brand_position	integer|null	Sentence number where brand first appears (1-based)
brand_recommendation	string	recommended, mentioned, or not_mentioned
brand_sentiment	string|null	positive, neutral, or negative
brand_context	string|null	1–2 sentences surrounding the brand mention
competitor_appearances	array	Per-competitor: mentioned, position, recommendation, sentiment, context
share_of_voice	float	Your brand's fraction of all brand+competitor mentions (0–1)
cited_sources	array	URLs cited as sources in the AI overview
overview_excerpt	string	First 300–550 chars of the AI response

Run Summary (KV store LAST_RUN_SUMMARY):

Field	Description
brand_visibility_score	0–100 composite score: mention rate (40%), recommendation rate (35%), position quality (15%), sentiment (10%)
overview_presence_score	0–100: % of queries that trigger rich AI overviews
brand_mention_rate	Overall fraction of checks where brand was found
brand_recommendation_rate	Fraction where brand is specifically recommended
avg_brand_position	Average sentence position when mentioned (lower = better)
sentiment_breakdown	{positive, neutral, negative} as fractions
share_of_voice	Map of brand + each competitor → their average SOV
platform_scores	Per-platform breakdown of all metrics
keyword_scores	Per-keyword breakdown of all metrics
competitor_threats	Competitors ranked by mention rate and recommendation rate
keywords_with_no_brand_mention	Keywords where brand is completely absent from overviews
keywords_where_competitor_dominates	Keywords where your SOV < 30%
recommendations	Up to 5 specific, actionable GEO recommendations

Pricing

• $0.096 per check (one keyword × one query × one platform)
• + $0.00005 per GB-second for compute (standard Apify rate)
• All AI API costs included — no API keys to bring, no OpenRouter or Gemini account needed

Cost examples:

Run scope	Checks	Cost
5 keywords × 3 queries × 4 platforms	60	~$5.76
10 keywords × 3 queries × 4 platforms	120	~$11.52
20 keywords × 2 queries × 2 platforms	80	~$7.68
Quick spot-check: 5 keywords × 1 query × 2 platforms	10	~$0.96

Platform Coverage

Platform	Model	Search method
Perplexity	Sonar	Native real-time web search with source citations
ChatGPT	GPT-4o Search Preview	OpenAI web search with inline citations
Claude	Claude Sonnet 4.5	web_search_20250305 live tool via OpenRouter
Gemini	Gemini 2.5 Flash	Native Google Search grounding (direct Google API)

All four platforms perform live web search — results reflect current AI overview content, not static training data.

Query Templates

The actor supports six purchase-intent query templates designed to trigger commercial AI overviews:

Template	Example query generated
best_of	"What are the best [keyword] solutions available in 2026?"
comparison	"How do the leading [keyword] platforms compare to each other?"
how_to_choose	"How should a business choose the right [keyword] solution?"
buyer_intent	"What [keyword] solution do most companies use and why?"
feature_query	"What features should I look for in a good [keyword] platform?"
versus	"[brand] vs alternatives for [keyword] — which do experts recommend?"

You can also provide customQueries with {keyword} and {brand} placeholders for fully custom questions.

AI-Agent & MCP Usage

This actor is designed as a first-class MCP tool for AI agents. In Claude, ChatGPT, or any MCP-compatible agent:

Tool: apify--ai-overviews-tracker

Input: {
  "brandName": "HubSpot",
  "keywords": ["best CRM for small business", "email marketing software"],
  "competitors": ["Salesforce", "Pipedrive", "ActiveCampaign"],
  "platforms": ["perplexity", "chatgpt"],
  "maxQueriesPerKeyword": 2
}

The actor returns structured JSON — brand visibility scores, competitor comparison, per-keyword breakdown — directly consumable by agents building GEO reports or competitive intelligence pipelines.

Input aliases for agents (all accepted equally):

• brandName → also accepts: brand, company, product, name
• keywords → also accepts: keyword, topics, searchTerms, queries
• competitors → also accepts: competitorBrands, rivals, alternatives

Usage Examples

Quick brand spot-check

{
  "brandName": "Notion",
  "keywords": ["best project management tools", "note-taking apps for teams"],
  "competitors": ["Obsidian", "Confluence", "Monday.com"],
  "platforms": ["perplexity", "chatgpt"],
  "maxQueriesPerKeyword": 1
}

Full GEO audit

{
  "brandName": "Stripe",
  "keywords": [
    "best payment processing software",
    "online payment gateway comparison",
    "payment API for developers",
    "ecommerce payment solutions",
    "SaaS billing software"
  ],
  "competitors": ["PayPal", "Braintree", "Square", "Adyen"],
  "platforms": ["perplexity", "chatgpt", "claude", "gemini"],
  "queryTemplates": ["best_of", "comparison", "how_to_choose", "versus"],
  "maxQueriesPerKeyword": 3,
  "responseFormat": "detailed",
  "webhookUrl": "https://hooks.zapier.com/hooks/catch/..."
}

Custom competitor comparison

{
  "brandName": "Figma",
  "keywords": ["UI design tools", "prototyping software"],
  "competitors": ["Sketch", "Adobe XD", "Canva"],
  "queryTemplates": ["versus", "best_of"],
  "customQueries": ["What design tool do product teams prefer for {keyword} in 2026?"],
  "maxQueriesPerKeyword": 2
}

How Brand Detection Works

The actor uses sentence-level analysis (not just keyword matching) to precisely identify brand mentions and context:

1. Mention detection: Splits the AI response into sentences, finds the first sentence containing your brand name (case-insensitive).
2. Position tracking: Records the sentence number (1-based) where the brand appears — earlier = AI sees you as a primary option.
3. Recommendation detection: Checks for recommendation phrases ("recommended", "top pick", "consider", "go with", "#1", "ideal for") within 200 characters of the brand mention.
4. Sentiment analysis: Measures positive vs negative signal words in the 2-3 sentence window around the brand mention.
5. Share of voice: Counts your brand's mentions vs all competitor mentions in the same overview to compute SOV (0–1).

How Share of Voice is Calculated

share_of_voice = brand_mentions / (brand_mentions + all_competitor_mentions) per check.

• 1.0 = only your brand mentioned in the overview (no competitors)
• 0.5 = your brand and one competitor mentioned equally
• 0.0 = brand absent from the overview

The run summary reports average SOV across all checks for each brand/competitor.

Webhook Integration

Send results to any HTTP endpoint when the run completes:

{
  "webhookUrl": "https://hooks.slack.com/services/...",
  "brandName": "Ahrefs",
  "keywords": ["SEO software", "backlink analysis tools"]
}

The POST payload is the full LAST_RUN_SUMMARY object — brand visibility scores, competitor threats, recommendations — ready for Slack notifications, Make.com workflows, or n8n automations.

Related Actors (AI Search Intelligence Suite)

Actor	What it tracks	Best for
AI Brand Monitor	Brand name mentions in informational AI answers	Brand/PR teams
AI Search Visibility Tracker	Domain cited as source in AI answers	SEO/AEO teams
AI Overviews Tracker (this actor)	Brand in commercial AI overview boxes	GEO/marketing teams

Run all three together for a complete AI search presence picture: reputation + source authority + commercial visibility.

Legal Disclaimer

This actor queries AI platforms (Perplexity, OpenAI, Anthropic, Google) through their public APIs and OpenRouter. Use is subject to each platform's Terms of Service. The actor is intended for legitimate brand monitoring and competitive intelligence purposes. Results reflect AI-generated content at the time of the run and may vary across subsequent runs. Apify and the actor developer are not affiliated with Perplexity, OpenAI, Anthropic, or Google.