Google AI Overview Scraper: AI answers, cited sources and brand tracking

Google AI Overview Scraper captures the AI Overview Google shows at the top of a search, for any list of queries. Each row has the full answer (plain text, Markdown and sections), every cited source with title, domain and URL, and a check of whether your domains are cited and your brands are mentioned. The organic top 10, People Also Ask and related searches from the same page come free. Any country, city, language or device, in bulk, with no proxies or browser.

• 🤖 Full AI Overview: plain text, Markdown with headings and lists, and sections that show which sources each part cites. AI Overviews that Google loads after the page are captured too.
• 🔗 Every cited source: deduplicated, numbered in citation order, with title, source name, domain, URL and snippet (often 10+ per answer).
• 🎯 AEO / GEO tracking built in: add your domains and brands; each row says whether you're cited, at which source position, your organic position, and which brands are mentioned or missing.
• 📄 The rest of the page, free: organic top 10, People Also Ask, related searches and the list of SERP features.
• 🌍 Countries, cities, languages, desktop or mobile.
• 💸 $2.50 per 1,000 queries (standard) or $6 per 1,000 (fast). Failed queries are free.

Sample output

Query	AI Overview words	Sources	Top cited domains
what is generative engine optimization	446	12	coursera.org, seerinteractive.com, jasper.ai
semrush vs ahrefs	288	10	stylefactoryproductions.com, onelittleweb.com, ecommercetrix.com
how to get a passport fast	313	7	travel.state.gov, photoaid.com, cookcountyclerkofcourt.org
best project management software for agencies	287	6	reddit.com, youtube.com, paymoapp.com

What data do you get?

One row per query:

Field	What it is
query, status	The search, and ai_overview, no_ai_overview or failed
has_ai_overview	Whether Google showed an AI Overview
ai_overview_text / ai_overview_markdown	The answer as plain text and as Markdown
ai_overview_sections	Each part of the answer with its title, text and the positions of the sources it cites
references	Every cited source: position, title, source, domain, url, snippet
reference_count, referenced_domains	Number of unique sources, and cited domains in order
target_domains, target_cited, target_first_reference_position	For your domains: cited or not, source positions, organic position
brands_mentioned, brands_not_mentioned	Which of your brands appear in the answer text
organic_results	Organic top 10: position, title, URL, domain, description
people_also_ask, related_searches, serp_features	The rest of the results page
ai_overview_is_async, google_url, location, language, device, checked_at	Context for every check

Two ready-made views: AI Overviews (one row per query) and Cited sources (one row per source, for counting which sites Google trusts).

How to scrape Google AI Overviews

1. Enter your search queries, one per line. Questions, comparisons and "best X" searches trigger AI Overviews most often.
2. Optional: add your domains and brands to check citations and mentions.
3. Pick the location (country or city), language and device.
4. Click Start, then export the results as JSON, CSV or Excel, or read them through the API.

Input example

{
    "queries": ["best crm for small business", "hubspot vs salesforce"],
    "targetDomains": ["hubspot.com"],
    "brands": ["HubSpot", "Salesforce", "Zoho"],
    "location": "United States",
    "language": "English",
    "device": "desktop",
    "mode": "standard"
}

Output example

{
    "query": "semrush vs ahrefs",
    "has_ai_overview": true,
    "status": "ai_overview",
    "ai_overview_text": "Both Ahrefs and Semrush are industry-leading digital marketing and SEO platforms, but they cater to slightly different core strengths.\nAhrefs\n- Best for: Pure organic SEO, backlink analysis, and managing multiple sites. …",
    "ai_overview_word_count": 288,
    "references": [
        {
            "position": 1,
            "title": "Ahrefs vs Semrush (2026) — Which Should You Use?",
            "source": "www.stylefactoryproductions.com",
            "domain": "stylefactoryproductions.com",
            "url": "https://www.stylefactoryproductions.com/blog/ahrefs-vs-semrush",
            "snippet": "Reasons to use Semrush over Ahrefs * Semrush is not just a content visibility tool. …"
        }
    ],
    "reference_count": 10,
    "referenced_domains": ["stylefactoryproductions.com", "onelittleweb.com", "ecommercetrix.com", "youtube.com", "ahrefs.com", "reddit.com", "drewgarrett.org", "seranking.com"],
    "target_domains": [
        {"domain": "semrush.com", "cited": false, "reference_positions": [], "organic_position": null},
        {"domain": "ahrefs.com", "cited": true, "reference_positions": [6], "organic_position": 4}
    ],
    "target_cited": true,
    "target_first_reference_position": 6,
    "brands_mentioned": ["Semrush", "Ahrefs"],
    "people_also_ask": ["Is Ahrefs still relevant?", "What are the top 3 SEO tools?"],
    "serp_features": ["ai_overview", "discussions_and_forums", "organic", "people_also_ask", "people_also_search", "perspectives", "video"],
    "location": "Chicago,Illinois,United States",
    "language": "English",
    "device": "mobile",
    "checked_at": "2026-10-23T02:16:25Z"
}

Shortened: references, organic_results and ai_overview_sections hold every item in a real run.

Pricing

Pay per query, billed by Apify. No start fee, no subscription.

Event	Price	What it is
Query checked (query-checked)	$0.0025 ($2.50 / 1,000)	One Google results page checked in standard mode, with or without an AI Overview
Query checked, fast (query-checked-fast)	$0.006 ($6 / 1,000)	The same, fetched live for the quickest answer

You're never charged for failed queries (they're saved with status: "failed" and the reason), the organic top 10, People Also Ask or related searches. Duplicate queries in one run are removed before checking.

You pay for queries that show no AI Overview, because the page was still checked and "no AI Overview here" is a result for tracking. To keep costs down, check queries that are likely to trigger one (questions, comparisons, "best X"). The run SUMMARY reports your AI Overview rate.

Examples: 1,000 queries cost $2.50 in standard mode. Tracking 200 queries every week costs about $2.17 a month. Set Maximum cost per run in the run options to cap your spend. Free Apify plan: up to 10 queries per run.

Standard or fast?

Both return the same data. Standard collects the whole batch together; in our Oct 22 tests a run of 8 queries finished in 24 seconds, and at busy times it can take a few minutes. If a query isn't ready after 7 minutes, the Actor fetches it live at no extra charge. Fast fetches every query live and in parallel, usually 5-25 seconds per query, and skips the queue entirely. It suits AI agents and one-off checks.

Failed queries

If Google or the data service can't return a page after several retries, the row has status: "failed", has_ai_overview: null and an error. Failed queries are not charged and are listed in the run SUMMARY (failedQueries), so you can re-run just those.

Use it through the API, on a schedule or with AI agents

One-link MCP setup: https://mcp.apify.com?tools=wrenfield-data/google-ai-overview-scraper (Claude, Cursor, VS Code or any MCP client). Set maxTotalChargeUsd on the run to cap what it can charge.

Run it from Python, JavaScript or any HTTP client with the Apify API, schedule it weekly to track AI Overview citations over time, or connect it to Make, Zapier, n8n and Google Sheets.

curl -X POST "https://api.apify.com/v2/acts/wrenfield-data~google-ai-overview-scraper/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"queries": ["best crm for small business"], "targetDomains": ["hubspot.com"], "mode": "fast"}'

FAQ

Why does a query have no AI Overview? Google decides per search, location and device, and it changes from day to day. Navigational searches ("facebook login") and very local or sensitive searches show AI Overviews less often. Those rows have status: "no_ai_overview" and still include the organic results.

Does it capture AI Overviews that load after the page? Yes. Google sometimes generates the AI Overview after the results load. The Actor waits for it, and ai_overview_is_async marks those rows.

How do I track whether Google's AI cites my site? Add your domain under Your domains. Each row shows cited, the source positions and your organic position. Schedule the run weekly and compare target_cited over time. Add competitors' domains too to see your share of citations.

Can I target a city or another language? Yes. Use a full Google location name such as London,England,United Kingdom, and any language Google supports (e.g. de, French).

Is this Google AI Mode? No. AI Overviews are the AI answers at the top of normal Google results. For AI Mode (the separate chat-style tab), ChatGPT, Perplexity and Gemini, see the related tools below.

Related tools

• AI Search Visibility Tracker: brand mentions, share of voice and citations across ChatGPT, Perplexity, Gemini, Google AI Mode and AI Overviews
• AI Answer Scraper: full answers and citations from ChatGPT, Perplexity, Gemini and Google AI Mode
• People Also Ask Scraper: PAA question trees up to 4 levels deep, plus related searches
• Competitor Keywords: every keyword a domain ranks for, to find queries where AI Overviews could cite you
• Keyword Search Volume: Google Keyword Planner volume, CPC and trends, to pick which queries to track

Feedback and support

Missing a field, hit an error, or need another option? Open an issue on the Issues tab with the run ID. We read every issue and reply there.

Acceptable use

• This Actor is an independent tool. It is not affiliated with, endorsed by or sponsored by Google or DataForSEO.
• Don't use the data to build a competing search engine, or otherwise to compete with or harm Google or the other search engines the data comes from.
• The Actor may not be used from, or for people or organizations in, Russia, Belarus, Cuba, Iran, Syria, Myanmar, North Korea, Crimea or the Russian-occupied territories of Ukraine.
• You are responsible for using the results lawfully, including privacy laws such as GDPR for any personal data they contain.