🤖 ChatGPT Search Scraper

🚀 Export ChatGPT Search answers in seconds. Send any list of questions and get back the full answer, every source ChatGPT cites, which sentence cites which source, the brands it highlights, its shopping cards and your own brand checks. In our cloud tests each answer took 6 to 14 seconds, with up to 9 cited sources and 17 websites searched per query.

This Actor asks ChatGPT your questions the way a logged-out visitor does on chatgpt.com, with the Web search tool switched on, and reads the answer straight from the data the ChatGPT web app renders: the citation pills, the "Sources" list, the product cards and the entity chips. Nothing is summarised or guessed. Each query runs in a fresh chat, so one question never colours the next.

Every row has 18 fields: the answer as plain text and as Markdown (tables and lists kept), how many websites ChatGPT searched, the cited sources with title, URL, domain and publisher, claim-level citations, the companies and products ChatGPT tagged as entities, shopping cards with price and merchant, and a visibility check for any brand or domain you list.

🎯 Target Audience	💡 Primary Use Cases
SEO and GEO specialists	Track which pages ChatGPT cites for your keywords
Brand and PR teams	See how ChatGPT describes your brand and your rivals
Content marketers	Find the sources that win AI citations in your niche
E-commerce teams	Watch which products and merchants ChatGPT recommends
Market researchers	Collect AI answers at scale for analysis
Agencies	Report AI search visibility to clients, query by query

📋 What the ChatGPT Search Scraper does

1. Takes a list of questions or keywords, one query per entry.
2. Opens a fresh logged-out ChatGPT chat for each query and turns on Web search (you can switch that off).
3. Waits for the complete answer and reads it from the page: text, Markdown, citation pills, the sources list, entity chips and product cards.
4. Strips ChatGPT's utm_source=chatgpt.com tag from every link so URLs match your own analytics.
5. Checks each brand or domain you list: named in the answer or not, how many times, where it first appears, and whether it is among the cited sources.
6. Writes one row per query. Failed queries are retried with a fresh session and never charged.

💡 Why it matters: more and more buyers ask ChatGPT before they ask Google. The answer they read names a handful of brands and cites a handful of pages. If your brand or your content is not in that answer, you are invisible at the moment of decision, and the only way to know is to ask the same questions ChatGPT's users ask, at scale, and keep the evidence.

🎬 Full Demo (🚧 Coming soon)

📊 Output

Each dataset row is one query and its ChatGPT answer. 18 columns per row.

Field	Type	Description
🔎 query	string	The question you sent
💬 answerText	string	The full answer as plain text
📝 answerMarkdown	string	The same answer in Markdown, with headings, lists, bold and tables
🔢 answerWordCount	number	Words in the answer
🌐 webSearchUsed	string	Yes when ChatGPT searched the web for this answer
🕸 websitesSearched	number	Websites ChatGPT reports it searched ("Searched 16 websites")
📚 sourcesCount	number	Cited sources
🏷 citedDomains	array	Unique domains of the cited sources, in citation order
🔗 sources	array	Cited sources: position, title, URL, domain, publisher
📌 citations	array	Each cited sentence of the answer with the sources that back it
🛒 productsCount	number	Shopping cards in the answer
🛍 products	array	Product cards: title, price, merchant, rating, review count, images, product ID
🏢 entitiesCount	number	Entity chips in the answer
🧩 entities	array	Companies, products and people ChatGPT tagged as entity chips, with its own category and description (only some answers have them)
🎯 trackedBrands	array	Per brand or domain you listed: mentioned, mention count, first position, cited, cited position
⏱ responseTimeSeconds	number	Seconds ChatGPT took to answer
🕒 scrapedAt	string	When the row was collected
❌ error	string	Populated only on failed rows

Product fields ChatGPT does not show for a card (often rating and reviews) come back as Not Disclosed. Queries without shopping cards or entity chips return empty arrays, never nulls.

Real sample records

A full row, with brand tracking for HubSpot and salesforce.com:

{
  "query": "What are the top 5 CRM tools",
  "answerText": "If you mean top CRM platforms in 2026, rather than a strict ranking, these five are among the most prominent and widely used:\n\n| CRM | Best suited for | Key strength |\n| Salesforce | Large/complex organizations | Deep customization, automation, integrations |\n| HubSpot CRM | SMBs and growing companies | Easy to use + strong sales/marketing integration |\n| Microsoft Dynamics 365 | Microsoft-centric businesses | CRM + Microsoft ecosystem integration |\n| Zoho CRM | Budget-conscious businesses | Broad features and relatively low cost |\n| Pipedrive | Sales-focused SMBs | Simple, visual sales pipeline |\n\nSalesforce remains the largest CRM vendor by market share, while current 2026 comparisons consistently include Salesforce, HubSpot, Microsoft Dynamics, Zoho, and Pipedrive among the leading platforms.\n\nHow I'd narrow it down\n\n- Want the most powerful/customizable: Salesforce\n- Want easy implementation and an all-in-one sales/marketing platform: HubSpot\n- Already use Microsoft 365/Teams: Dynamics 365\n- Want lots of functionality for less money: Zoho\n- Primarily need a straightforward sales pipeline: Pipedrive\n\nIf you tell me your company size, number of salespeople, and approximate monthly CRM budget, I can give you a side-by-side comparison of the five, including pricing, AI features, automations, integrations, and which one fits your use case.",
  "answerMarkdown": "If you mean **top CRM platforms in 2026**, rather than a strict ranking, these five are among the most prominent and widely used:\n\n| CRM | Best suited for | Key strength |\n| --- | --- | --- |\n| **Salesforce** | Large/complex organizations | Deep customization, automation, integrations |\n| **HubSpot CRM** | SMBs and growing companies | Easy to use + strong sales/marketing integration |\n| **Microsoft Dynamics 365** | Microsoft-centric businesses | CRM + Microsoft ecosystem integration |\n| **Zoho CRM** | Budget-conscious businesses | Broad features and relatively low cost |\n| **Pipedrive** | Sales-focused SMBs | Simple, visual sales pipeline |\n\nSalesforce remains the largest CRM vendor by market share, while current 2026 comparisons consistently include Salesforce, HubSpot, Microsoft Dynamics, Zoho, and Pipedrive among the leading platforms.\n\n### How I'd narrow it down\n\n- **Want the most powerful/customizable:** Salesforce\n- **Want easy implementation and an all-in-one sales/marketing platform:** HubSpot\n- **Already use Microsoft 365/Teams:** Dynamics 365\n- **Want lots of functionality for less money:** Zoho\n- **Primarily need a straightforward sales pipeline:** Pipedrive\n\nIf you tell me **your company size, number of salespeople, and approximate monthly CRM budget**, I can give you a side-by-side comparison of the five, including **pricing, AI features, automations, integrations, and which one fits your use case**.",
  "answerWordCount": 209,
  "webSearchUsed": "Yes",
  "websitesSearched": 16,
  "sourcesCount": 3,
  "citedDomains": [
    "salesforce.com",
    "gartner.com",
    "softwareadvice.com"
  ],
  "sources": [
    {
      "position": 1,
      "title": "Salesforce Ranked #1 CRM Provider for 13th Year - Salesforce",
      "url": "https://www.salesforce.com/news/stories/idc-crm-market-share-ranking-2026/",
      "domain": "salesforce.com",
      "publisher": "Salesforce"
    },
    {
      "position": 2,
      "title": "Gartner Magic Quadrant for CRM Sales Platforms",
      "url": "https://www.gartner.com/en/documents/8230961",
      "domain": "gartner.com",
      "publisher": "Gartner"
    },
    {
      "position": 3,
      "title": "25 Best CRM Software - 2026 Reviews & Pricing",
      "url": "https://www.softwareadvice.com/crm/",
      "domain": "softwareadvice.com",
      "publisher": "Software Advice"
    }
  ],
  "citations": [
    {
      "text": "Salesforce remains the largest CRM vendor by market share, while current 2026 comparisons consistently include Salesforce, HubSpot, Microsoft Dynamics, Zoho, and Pipedrive among the leading platforms.",
      "sources": [
        {
          "title": "Salesforce Ranked #1 CRM Provider for 13th Year - Salesforce",
          "url": "https://www.salesforce.com/news/stories/idc-crm-market-share-ranking-2026/",
          "domain": "salesforce.com",
          "publisher": "Salesforce"
        },
        {
          "title": "Gartner Magic Quadrant for CRM Sales Platforms",
          "url": "https://www.gartner.com/en/documents/8230961",
          "domain": "gartner.com",
          "publisher": "Gartner"
        },
        {
          "title": "25 Best CRM Software - 2026 Reviews & Pricing",
          "url": "https://www.softwareadvice.com/crm/",
          "domain": "softwareadvice.com",
          "publisher": "Software Advice"
        }
      ]
    }
  ],
  "productsCount": 0,
  "products": [],
  "entitiesCount": 0,
  "entities": [],
  "trackedBrands": [
    {
      "brand": "HubSpot",
      "mentioned": "Yes",
      "mentionCount": 3,
      "firstMentionAtChar": 263,
      "cited": "No",
      "citedPosition": "N/A"
    },
    {
      "brand": "salesforce.com",
      "mentioned": "Yes",
      "mentionCount": 4,
      "firstMentionAtChar": 171,
      "cited": "Yes",
      "citedPosition": 1
    }
  ],
  "responseTimeSeconds": 6.8,
  "scrapedAt": "2026-09-24T20:49:03.170Z",
  "error": null
}

A shopping query, arrays shortened to the first entries:

{
  "query": "best running shoes for flat feet 2026",
  "answerWordCount": 331,
  "webSearchUsed": "Yes",
  "websitesSearched": 4,
  "sourcesCount": 4,
  "citedDomains": [
    "runnersworld.com",
    "runrepeat.com"
  ],
  "sources": [
    {
      "position": 1,
      "title": "The 8 Best Running Shoes for Flat Feet in 2026 - Shoes for Flat-Footed Runners",
      "url": "https://www.runnersworld.com/gear/a25750345/running-shoes-flat-feet/",
      "domain": "runnersworld.com",
      "publisher": "Runner's World"
    },
    {
      "position": 2,
      "title": "70+ Stability Running Shoe Reviews | RunRepeat",
      "url": "https://runrepeat.com/catalog/stability-running-shoes",
      "domain": "runrepeat.com",
      "publisher": "#1 Athletic Shoe Review Site"
    }
  ],
  "citations": [
    {
      "text": "Flat feet don't automatically mean you need aggressive arch support. Current guidance emphasizes comfort, fit, and how your foot actually behaves while running rather than simply selecting the shoe with the biggest arch bump.",
      "sources": [
        {
          "title": "The 8 Best Running Shoes for Flat Feet in 2026 - Shoes for Flat-Footed Runners",
          "url": "https://www.runnersworld.com/gear/a25750345/running-shoes-flat-feet/",
          "domain": "runnersworld.com",
          "publisher": "Runner's World"
        }
      ]
    }
  ],
  "productsCount": 5,
  "products": [
    {
      "position": 1,
      "imageUrl": "https://images.openai.com/products/v1/40/4a/404adcce6fb3bc308f6b649b774cce6a2f7ac086b18c6ed106758d295bee5cdb.jpg",
      "title": "Brooks Adrenaline GTS 25",
      "price": "$155.00",
      "merchant": "Nordstrom",
      "rating": "Not Disclosed",
      "reviewCount": "Not Disclosed",
      "productId": "A4721147",
      "imageUrls": [
        "https://images.openai.com/products/v1/40/4a/404adcce6fb3bc308f6b649b774cce6a2f7ac086b18c6ed106758d295bee5cdb.jpg"
      ]
    },
    {
      "position": 2,
      "imageUrl": "https://images.openai.com/products/v1/28/5f/285fed136685b06a3174ee1dc0d1feb49a854eac3f6de5bffa8b955b0a8fd2c6.jpg",
      "title": "ASICS GEL-KAYANO 32",
      "price": "$124.95",
      "merchant": "Nordstrom",
      "rating": "Not Disclosed",
      "reviewCount": "Not Disclosed",
      "productId": "A9371865",
      "imageUrls": [
        "https://images.openai.com/products/v1/28/5f/285fed136685b06a3174ee1dc0d1feb49a854eac3f6de5bffa8b955b0a8fd2c6.jpg"
      ]
    }
  ],
  "entitiesCount": 0,
  "entities": [],
  "responseTimeSeconds": 10.4,
  "scrapedAt": "2026-09-24T20:49:21.361Z",
  "error": null
}

The entity chips ChatGPT attached to a CRM answer (other fields omitted):

{
  "query": "What are the top 5 CRM tools",
  "sourcesCount": 4,
  "citedDomains": [
    "technologyadvice.com",
    "salesforce.com",
    "g2.com",
    "gartner.com"
  ],
  "entitiesCount": 5,
  "entities": [
    {
      "position": 1,
      "name": "Salesforce",
      "category": "company",
      "description": "CRM software company"
    },
    {
      "position": 2,
      "name": "HubSpot",
      "category": "company",
      "description": "CRM and marketing software company"
    },
    {
      "position": 3,
      "name": "Zoho",
      "category": "company",
      "description": "business software company"
    },
    {
      "position": 4,
      "name": "Microsoft",
      "category": "company",
      "description": "technology company"
    },
    {
      "position": 5,
      "name": "Pipedrive",
      "category": "company",
      "description": "CRM software company"
    }
  ],
  "scrapedAt": "2026-09-24T20:45:35.574Z",
  "error": null
}

✨ Why choose this Actor

• Claim-level citations. citations pairs each cited sentence of the answer with the exact sources behind it, so you see not only that a page was cited but what ChatGPT used it to say.
• Built-in brand visibility. List your brand, your domain and your competitors in Brands or domains to track and every row says whether each one is named, how often, how early, and whether it is cited. No spreadsheet formulas needed.
• Shopping cards included. Product recommendations come with price, merchant, rating, review count, product ID and images.
• Answer text and Markdown. Tables, lists and headings survive in answerMarkdown, ready for an LLM pipeline or a report. answerText is the same answer flattened.
• Entity chips when ChatGPT shows them. Some answers tag companies or products as clickable entities with a category and a one-line description; entities keeps them in order. ChatGPT shows these chips only in some answers (3 of about 30 in our tests), so treat the column as a bonus, not a guarantee.
• Clean URLs. The utm_source=chatgpt.com tag is removed from every source URL.
• Fresh chat per query. No conversation memory leaks from one question into the next.
• Charged only for answers. Queries that fail after retries are written as error rows and are never billed.

📈 How it compares to alternatives

	This Actor	Typical ChatGPT search scrapers	Checking by hand
Full answer text	Yes, plain text and Markdown	Plain text	Copy and paste
Cited sources with publisher and domain	Yes	Title and URL	Manual
Which sentence cites which source	Yes	No	Manual
Entity chips (when ChatGPT shows them)	Yes	No	Manual
Shopping cards with price and merchant	Yes	Some	Manual
Brand and domain visibility checks	Built in	No, separate tools	Manual
The follow-up searches ChatGPT ran	No, see below	Some	No
Needs your OpenAI account	No	Varies	Yes

ChatGPT does not show logged-out visitors the list of follow-up searches it ran or the pages it read but did not cite. The Actor reports the count ChatGPT displays ("Searched 16 websites") in websitesSearched, and the cited pages in full.

🚀 How to use

1. Create a free Apify account. New accounts include $5 of free platform credit, which is plenty to try this out: console.apify.com/sign-up
2. Open the Actor and go to the Input tab.
3. Add your questions to Search queries, one per entry. Write them the way your customers would ask ChatGPT.
4. Set Max Items to the number of queries to run.
5. Optionally list brands or domains in Brands or domains to track, for example your brand and three competitors.
6. Click Start. The log shows the words, sources, citations and products found for each query as it goes.
7. Open the Dataset tab and download as CSV, Excel, JSON or XML, or pull it from the API.

{
  "queries": ["What are the top 5 CRM tools", "best CRM for small business"],
  "maxItems": 2,
  "trackBrands": ["HubSpot", "salesforce.com", "Pipedrive"]
}

💼 Business use cases

Generative engine optimization (GEO) tracking

Build a list of the questions your buyers ask, run it weekly, and chart trackedBrands over time: how often ChatGPT names you, how early in the answer, and whether your own domain is among the cited sources. citedDomains shows which publishers you need to be featured on to get there.

Competitive intelligence

Put your brand and your top competitors in Brands or domains to track and run category questions ("best X for Y"). The rows show who ChatGPT recommends first, who it cites, and, when ChatGPT shows entity chips, the one-line description it attaches to each company in entities.

Content and PR strategy

citations shows the exact sentences ChatGPT backs with each source. Aggregate them across a keyword set and you get the list of review sites, publishers and pages that shape AI answers in your niche, and what those answers borrow from them.

E-commerce and product visibility

For shopping questions, products lists the cards ChatGPT shows with price and merchant. Track which of your products appear, at what price, and which retailers ChatGPT sends buyers to.

🔌 Automating ChatGPT Search Scraper

• Make and Zapier: run the Actor on a schedule and push new rows into a sheet or a dashboard.
• Slack: post an alert when a tracked brand drops out of an answer or a competitor appears.
• Airbyte: sync datasets into Snowflake, BigQuery or Postgres for long-term AI visibility reporting.
• GitHub: trigger runs from Actions and keep versioned snapshots of how answers change.
• Google Drive: save a CSV or Google Sheet to a shared folder after every run.
• API and webhooks: every run emits a dataset ID; subscribe to the run succeeded webhook and pull the rows into any system.

🌟 Beyond business use cases

• Research: study how AI search engines select and attribute sources, and how answers drift over weeks.
• Personal: compare what ChatGPT recommends for a purchase before you buy.
• Non-profit: check how ChatGPT describes your cause, and which sources it trusts on the topic.
• Experimentation: answerMarkdown plus citations is a clean dataset for evaluating grounding and citation quality.

🤖 Ask an AI assistant about this scraper

Paste this into ChatGPT, Claude or any assistant to get help designing your run:

I am using the ParseForge ChatGPT Search Scraper on Apify. It sends my queries to ChatGPT (logged out, web search on) and returns one row per query with 18 fields: query, answerText, answerMarkdown, answerWordCount, webSearchUsed, websitesSearched, sourcesCount, citedDomains, sources (position, title, url, domain, publisher), citations (cited sentence plus its sources), productsCount, products (title, price, merchant, rating, reviewCount, productId, images), entitiesCount, entities (name, category, description), trackedBrands (brand, mentioned, mentionCount, firstMentionAtChar, cited, citedPosition), responseTimeSeconds and scrapedAt. Help me design a query list and a brand list to answer this question: [your question here].

❓ Frequently Asked Questions

🔐 Do I need an OpenAI or ChatGPT account? No. The Actor uses ChatGPT the way a logged-out visitor does on chatgpt.com. You never hand over credentials or an API key.

🌐 Does every answer use web search? With Force web search on (the default), the Web search tool is switched on for every query and answers come with citations. For a few purely creative prompts ChatGPT may still answer without searching; webSearchUsed tells you which rows were grounded. Switch the option off to see what ChatGPT answers on its own.

🔎 Can I get the follow-up searches ChatGPT ran? No. ChatGPT does not show that list to logged-out visitors. You get the number of websites it reports searching in websitesSearched, and every source it cites in full.

📚 Why are there fewer sources than websites searched? ChatGPT reads more pages than it cites. sources lists what it cites in the answer; websitesSearched is the number it reports having looked at.

🎯 How does brand tracking match my brands? Names are matched case-insensitively in the answer text. A domain like salesforce.com is matched in the text by its name ("Salesforce") and against the cited sources by domain, so you learn both whether the brand is named and whether its site is cited.

🧩 What are entities? In some answers ChatGPT tags companies, products and people as clickable entity chips with its own category and description. The Actor returns them in order in entities. ChatGPT decides when to show them: in our tests about 3 answers in 30 had them, and the other rows return an empty array.

🛒 What currency are product prices in? The currency ChatGPT shows, which follows the country the question is asked from. Without a proxy the platform asks from the United States and prices come in US dollars. Turn on a residential proxy with a country to see local prices.

🛡 Do I need a proxy? No. The Actor runs without one by default, and without a proxy all 14 queries in our last cloud tests answered on the first try. A residential proxy is only useful to ask from a specific country; in our tests answers from residential exits were also shorter (70 to 190 words against 190 to 610 without a proxy).

⏱ How long does a run take? About 6 to 14 seconds per query in our cloud tests, plus a few seconds to start.

🔁 What happens when a query fails? It is retried up to four times with a fresh browser session. If it still fails, the row carries an error and is not charged.

🧠 Is each query independent? Yes. Every query runs in a brand-new chat, so earlier questions never influence later answers.

📥 What export formats are supported? CSV, Excel, JSON, XML, plus direct API access and integrations with Make, Zapier, Airbyte, Slack, Google Drive and more.

⚖️ Is this legal? The Actor collects the answers ChatGPT gives to any logged-out visitor. You remain responsible for how you use the data, including compliance with OpenAI's terms and applicable law.

🔌 Integrate with any app

Every run writes to an Apify dataset reachable through a REST API, so the output drops into whatever you already use. Native integrations cover Make, Zapier, Airbyte, Slack, Google Drive, GitHub, Google Sheets and webhooks, and the API covers everything else.

🔗 Recommended Actors

• Google Search Scraper - 23 Fields, Organic & Paid - compare ChatGPT's sources with Google's rankings for the same keywords.
• Bing Search Results Scraper - the index behind much of AI search.
• Google Shopping Scraper - Product Prices & Offers - check the prices ChatGPT's product cards quote.
• Similarweb Traffic Scraper - size up the domains ChatGPT cites.
• Reddit Scraper & Search API: Posts, Users, Subreddits - see what the communities ChatGPT reads are saying.

💡 Pro Tip: browse the complete ParseForge collection.

🆘 Need Help? Open our contact form

⚠️ Disclaimer: independent tool, not affiliated with OpenAI or ChatGPT; only publicly available data.