ChatGPT Search Scraper: Brand Mentions & Citations per Prompt
Pricing
from $5.60 / 1,000 prompt answereds
ChatGPT Search Scraper: Brand Mentions & Citations per Prompt
Scrape ChatGPT search answers for your prompt list, one row per prompt: answer in markdown, pages it cites, pages it searched, fan-out queries, and true/false flags showing if your brand and each competitor is mentioned or cited. 191 countries. $8 per 1,000 answers; failed prompts are free.
Pricing
from $5.60 / 1,000 prompt answereds
Rating
0.0
(0)
Developer
Tenfold Fleet
Maintained by CommunityActor stats
0
Bookmarked
3
Total users
2
Monthly active users
a day ago
Last modified
Features
Share
Independent tool. Not affiliated with ChatGPT or OpenAI.
What does ChatGPT Search Scraper do?
ChatGPT Search Scraper takes a list of prompts, puts each one to ChatGPT with web search switched on, and hands back a single row per prompt. That row holds the full answer in markdown, the sources ChatGPT cites, every page it looked at while searching, the fan-out queries it ran, and the brands it names. Give it your brand and your competitors and each row also says, in plain true/false fields, whether the answer mentions your brand, whether your site is one of the cited sources, and where you stand among the brands named.
Use it as a ChatGPT scraper API for GEO and AI-visibility work (generative engine optimization): run the same prompt list every week and track which answers name you, which websites ChatGPT trusts, and where a competitor is recommended instead of you. Answers are available for 191 countries and 128 languages. No ChatGPT account, API key, proxy or browser is needed on your side.
You pay $8 per 1,000 answered prompts in standard mode. Failed and empty answers are free.
Why use it for ChatGPT brand mentions and citations?
- AI visibility tracking: see on which prompts ChatGPT names your brand, and how early in the answer.
- Citation tracking: see whether your website is a source behind the answer, at which position, and with which URL.
- Competitor benchmarking: one row lists which competitors are mentioned and which are cited, so you can find the prompts where they win the answer.
- Content planning: read the fan-out queries and the cited pages to learn what ChatGPT searches for and which pages it relies on.
- Client reporting: export a CSV or Excel sheet per week and chart mention share and citation share.
- Automation and AI agents: call it from the API, Make, Zapier, n8n or the Apify MCP server.
What makes it different?
- The matching is done for you. Each prompt is one flat row with
brandMentioned,brandCited,brandPosition,competitorsMentionedandcompetitorsCited. You do not have to parse the answer or the source list yourself. - Cited sources and searched pages are kept apart.
sourcesholds only the pages ChatGPT actually cited;searchResultsholds everything it retrieved, used or not. - You only pay for answers. A prompt that fails, comes back empty or is not ready in time is returned as a free row with
erroranderrorCode, so your list always comes back complete.
What data can ChatGPT Search Scraper extract?
| Field | Description |
|---|---|
prompt | The prompt you sent |
answerMarkdown | ChatGPT's full answer in markdown (headings, lists, tables, links; image thumbnails removed) |
sources | Pages ChatGPT cited, in order: position, title, url, domain, snippet, sourceName, publishedAt |
sourceCount, citedDomains | Number of cited sources and the unique domains among them |
searchResults | Every page ChatGPT retrieved while searching: position, title, url, domain, description |
fanOutQueries | The web searches ChatGPT ran to answer the prompt |
brandEntities | Brands, products and companies ChatGPT names: name, category, urls |
brandMentioned | true if the answer names your brand |
brandCited | true if your domain is among the cited sources |
brandPosition | Your place among the brands named in the answer (1 = named first) |
brandCitationPosition, brandCitedUrl | Position and URL of your first cited page |
competitorsMentioned, competitorsCited | Which of your competitors the answer names and cites |
competitors | The same detail per competitor: mentioned, position, cited, citationPosition, citedUrl |
model | ChatGPT model that answered |
location, language, mode, checkedAt | Run settings and the time of the answer |
error, errorCode | null on success; the reason on a free, not-answered row |
How to scrape ChatGPT search answers
- Open the Actor and click Try for free.
- Paste your prompts, one per line (each up to 2,000 characters).
- Optional: enter your brand as a name, a domain or both (
HubSpot, hubspot.com) and add up to 20 competitors the same way. - Pick the country and language. Leave Force web search on to get cited sources.
- Click Start. A short list is usually ready in one to three minutes.
- Download the results as JSON, CSV, Excel or HTML, or read them through the API.
How much does it cost to scrape ChatGPT?
Pricing is pay per event, so you pay only for prompts that come back with an answer.
| Event | Price | Per 1,000 |
|---|---|---|
| Prompt answered (standard mode) | $0.008 | $8 |
| Prompt answered (fast mode) | $0.015 | $15 |
| Actor start | $0.00005 per run |
Failed prompts, empty answers, prompts not ready in time and skipped prompts are free. Checking 500 prompts every week in standard mode costs about $4 a week.
Standard or fast? Standard mode queues your prompts and collects the answers together; a short list is usually done in one to three minutes (10 prompts took about one minute in our test). A list over 100 prompts is sent in two steps: the first 100, then the rest as soon as the first answer is back. Fast mode asks ChatGPT directly with no queue, five prompts at a time, and usually answers each prompt within a minute. Use fast mode for one-off questions from an agent or a live workflow, and standard mode for tracking lists.
Free Apify plan: runs on the free plan answer the first 5 prompts per run in standard mode; further prompts come back as free rows marked as not answered. Every answer costs the developer a data fee, so this limit is set by the Actor developer, not by Apify. Any paid Apify plan runs every prompt and unlocks fast mode.
Input
{"prompts": ["best crm software for small business","is hubspot worth it for a 5 person team?"],"brand": "HubSpot, hubspot.com","competitors": ["Zoho, zoho.com", "Salesforce, salesforce.com"],"location": "United States","language": "en","forceWebSearch": true,"mode": "standard","maxPrompts": 50}
prompts(required): the questions to ask. Each prompt is a fresh, signed-out conversation, so answers do not depend on any chat history.brand: a name, a domain, or both separated by a comma. Give the domain for exact citation matching; subdomains count.competitors: up to 20, each written likebrand.location,language: a country name or 2-letter code, and a language code or name.forceWebSearch: on by default. Turn it off to see what ChatGPT says from memory, without sources.mode:standard(default) orfast.maxPrompts: safety cap per run (default 50, up to 1,000).
Output

One row per prompt (shortened here). You can download the dataset as JSON, CSV, Excel, XML or HTML.
{"prompt": "best crm software for small business","location": "United States","language": "en","mode": "standard","model": "gpt-6","answerMarkdown": "# Best CRM software for small businesses (2026)\n\nAfter comparing current CRM reviews, pricing and features, my top recommendation is **HubSpot for most small businesses**. ...","sourceCount": 10,"citedDomains": ["hubspot.com", "zoho.com", "pipedrive.com", "salesforce.com"],"brand": "HubSpot","brandMentioned": true,"brandCited": true,"brandPosition": 1,"brandCitationPosition": 1,"brandCitedUrl": "https://www.hubspot.com/products/crm","competitorsMentioned": ["Zoho", "Salesforce"],"competitorsCited": ["Zoho", "Salesforce"],"competitors": [{ "name": "Zoho", "domain": "zoho.com", "mentioned": true, "position": 2, "cited": true, "citationPosition": 2, "citedUrl": "https://www.zoho.com/crm/" }],"sources": [{ "position": 1, "title": "Visit HubSpot", "url": "https://www.hubspot.com/products/crm", "domain": "hubspot.com", "snippet": null, "sourceName": null, "publishedAt": null }],"searchResults": [{ "position": 1, "title": "Best CRM for Small Businesses (2026): 7 Picks", "url": "https://www.example.com/best-crm", "domain": "example.com", "description": "..." }],"fanOutQueries": ["best CRM software for small business 2026 comparison pricing features"],"brandEntities": [{ "name": "HubSpot CRM", "category": "software", "urls": [] }],"chatgptUrl": "https://chatgpt.com/?prompt=best%20crm%20software%20for%20small%20business&hints=search","checkedAt": "2026-11-16T04:18:38.000Z","error": null,"errorCode": null}
A prompt that could not be answered is a free row:
{"prompt": "…","location": "United States","language": "en","mode": "standard","error": "ChatGPT returned an empty answer (not charged)","errorCode": "EMPTY_ANSWER"}
errorCode is one of INVALID_PROMPT, TASK_FAILED, EMPTY_ANSWER, NOT_READY, MAX_PROMPTS, MAX_CHARGE, FREE_PLAN_LIMIT, TIMEOUT or SOURCE_UNAVAILABLE.
Tips
- Write prompts the way customers ask. "best payroll software for a 10 person company" tells you more than "payroll software".
- Track the same list on a schedule. ChatGPT's answers change from day to day; a weekly run of a fixed prompt list shows the trend. Use Apify Schedules.
- Enter both the brand name and the domain. The name drives
brandMentioned, the domain drivesbrandCited. With a name only, a source counts as yours when the name is part of its host name. brandPositionis about order, not ranking quality. It counts how many other brands ChatGPT names before yours in the answer.- Check one country at a time. Answers and cited sources differ by country; run the list once per market.
- Set a max cost per run if you want a hard spending cap. Prompts beyond the cap are returned as free rows.
FAQ
Do I need a ChatGPT account or an OpenAI API key?
No. The Actor needs no login, cookie or key from you. Each prompt is answered in a fresh, signed-out ChatGPT session with web search.
Is this the same answer I would see in ChatGPT?
It is the answer the ChatGPT web app gives a signed-out visitor in the chosen country at that moment. ChatGPT is not deterministic: asking again can change the wording, the brands named and the sources. That is why tracking over several runs is more reliable than a single check.
Why does an answer have no sources?
Even with web search forced, ChatGPT sometimes answers without citing pages, mostly for simple factual or conversational prompts. Such a row is a valid answer (sourceCount: 0) and is billed like any other.
How long does a run take?
In standard mode a short list is usually finished in one to three minutes, because the prompts are queued together. A list over 100 prompts takes one step longer: the first 100 go out, and the rest follow once the first answer has arrived. If an answer is not ready after 10 minutes the prompt is returned as a free NOT_READY row. Fast mode usually answers each prompt within a minute.
How is brandMentioned decided?
The answer text is searched for your brand name as a whole word, ignoring case, and the brands ChatGPT itself lists are checked too. Text inside image captions and link addresses is ignored.
Can I use it from an AI agent or through the API?
Yes. Call the Actor through the Apify API or the Apify MCP server with a prompts array, then read the dataset. For a single question from an agent, mode: "fast" gives the quickest reply.
Is scraping ChatGPT answers legal?
Every row is an answer anybody can get from ChatGPT without signing in. No account, private chat or paid feature is touched. What you do with the output is up to you, and that includes text from other websites quoted inside an answer. If unsure, consult your lawyer.
Disclaimer
This Actor is built and run independently: OpenAI has no part in it and does not endorse or sponsor it. "ChatGPT" is a trademark of its owner and is used here only to describe what the tool reads.
More tools from Tenfold Fleet
| Actor | Output |
|---|---|
| Brand Mentions Monitor - Bluesky, Hacker News & Mastodon | mention |
| Amazon Search Volume - Keyword Demand for Sellers & PPC | keyword with amazon volume |
| ASO Keyword Rank Tracker: iOS & Android App Positions | keyword check |
| Backlinks Checker - Referring Domains, Anchors & Domain Rank | backlink |
| Bing Search Volume: Microsoft Ads Keyword Data in Bulk | bing keyword batch |
| Google AI Overviews Checker - Citations & Domain Tracker | keyword checked |
| Google Images API - Image Search Results in 90+ Countries | image result |
| Google Keyword Planner Alternative - Ideas, Bids & Volume | keyword idea |
| Google Maps Rank Tracker - Geo-Grid Rank Changes & Competitors | grid point checked |
| Google Trends Scraper - Search Interest by Region and Over Time | keyword |
| Keyword Ideas & Related Keywords - Difficulty, Volume & CPC | keyword idea |
| Keyword Search Volume Checker - Bulk CPC & Keyword Difficulty | keyword with data |
All 122 tools: apify.com/tenfoldfleet. If this Actor saves you time, save it (the bookmark on its Store page) and leave a review: both help other people find it. Found a bug or need another field? Tell us on the Issues tab of this page.