Facebook Search Scraper - Pages, Posts, Videos & Groups

Facebook Search Scraper searches Facebook by keyword and exports public pages, posts, videos and reels, and groups from one run. It can also return people, places and events. Every row keeps the keyword that found it, the result type and its rank, so you can track brands, campaigns and topics over time.

No Facebook account, cookies or login are needed. The Actor reads Facebook's own logged-out search results, not Google site: results.

What does Facebook Search Scraper do?

• Searches Facebook for each keyword you enter, one tab at a time: Pages, Posts, Videos, Groups, People, Places, Events.
• Saves one structured row per result card, with a shared set of columns across all result types.
• Pages: name, Facebook ID, URL, description snippet, website, verified badge, profile picture.
• Posts: text, permalink, author, group the post was shared in, exact publish time, reactions, comments, shares, attachment type and image.
• Videos and reels: title or caption, video ID, URL, creator, publish time, views, reactions, comments, shares, duration, thumbnail.
• Groups: the public groups that matching posts were shared in, with group ID, name, URL and how many matching posts were found there.
• Optional Most recent first ordering and a date window for posts and videos, for daily monitoring.
• Removes the unrelated popular posts Facebook adds to thin searches, so a narrow keyword does not return filler.

Who is it for?

• Brand and social listening teams that check daily what is being posted about a brand, product or campaign hashtag.
• Agencies that need a list of pages, creators and groups around a niche before outreach or influencer research.
• Market researchers comparing how competitors appear in Facebook search.
• Lead generation teams collecting business pages and places with websites for a keyword and city, for example dentist chicago.
• Community managers looking for groups where a topic is discussed.
• Data teams feeding Facebook search results into dashboards, spreadsheets or AI workflows on a schedule.

Why use this Facebook search scraper?

• Several result types in one run. Pick any mix of pages, posts, videos, groups, people, places and events. You don't need a separate Actor per tab.
• No login. Nothing to set up, no account at risk of being restricted.
• Engagement numbers on posts and videos. Reactions, comments, shares and video views are included whenever Facebook shows them on the result card. Some video cards have no counts, and those fields are then null.
• Exact dates. Posts and most videos have an ISO timestamp, not just "3 days ago". Some native video cards have no date (publishedAt is null).
• Relevant results only. By default, posts and videos must mention every word of your keyword. Facebook's filler content is dropped and not charged.
• Recency and date filters built for repeat monitoring jobs. Dated posts and videos outside the window are dropped; native videos without a date are kept and have publishedAt: null.
• Fast HTTP extraction. No browser and low memory. Every result is charged before it is saved, and the run never saves more than Max total results or your maximum cost per run allows.

What data can you extract from Facebook search?

Field	Pages / places	People	Events	Posts	Videos	Groups
facebookId	✅	✅	✅	✅ post ID	✅ video ID	✅ group ID
name	✅	✅	✅			✅
text (snippet, post text, video title)	✅			✅	✅
url	✅	✅	✅	✅	✅	✅
authorName, authorId, authorUrl				✅	✅
groupId, groupName, groupUrl				✅ when posted in a group		✅
publishedAt / publishedAtText				✅	most
reactionCount, commentCount, shareCount				✅ when shown	✅ when shown
viewCount, durationSeconds				some	✅
mediaType, imageUrl	picture	picture	picture	✅	✅ thumbnail
isVerified	✅	✅			some
websiteUrl	✅ when shown
location		✅ "Lives in …"
matchingPostCount						✅

Every row also has query, resultType, rank, searchModule and scrapedAt. Fields that do not apply to a result type are null.

How much does it cost to scrape Facebook search results?

This Actor uses pay-per-event pricing. You pay a small fee when a run starts and then for each result saved to the dataset. Platform usage is included.

Event	Price
Run started (once per run)	$0.002
Search result, Free and Starter plans	$0.00313
Search result, Scale plan	$0.002817
Search result, Business plan and above	$0.002504

Examples at the Starter plan price:

• 1 keyword × 4 result types × 25 results = 100 results → about $0.32.
• 10 brand keywords monitored daily with 20 recent posts each = 200 results → about $0.63 per day.
• 1,000 results → about $3.13.

Apify's free plan includes $5 of monthly credit, enough for about 1,500 results. The $0.002 start fee is charged once for every run with valid input, even when Facebook returns nothing. Result fees are charged only for rows saved to the dataset, so rows dropped by the keyword filter, empty searches and failed requests have no result fee. Invalid input fails before any charge.

To cap the cost of a run, use Max total results or set a maximum cost per run in the run options. The Actor charges each result before saving it and stops cleanly when either limit is reached, so you never get more rows than you allowed and never get rows that were not charged. The status message then says how many results were delivered and why the run stopped.

How to scrape Facebook search results

1. Open the Actor and enter one or more Search keywords, for example nike, coffee shop chicago or #blackfriday.
2. Choose the Result types you need. Pages, posts, videos and groups are selected by default.
3. Set Max results per keyword and type, for example 20.
4. Optional: turn on Most recent first and set a date window for posts and videos.
5. Click Start and wait for the run to finish. A few hundred results usually take a minute or two.
6. Download the results as JSON, CSV, Excel or XML, or open the Posts and videos and Pages, groups, people, places and events views.

Input parameters

Parameter	Type	Default	Description
searchQueries	array of strings	required	Keywords, brand names, hashtags or phrases. Up to 100 per run.
resultTypes	array	pages, posts, videos, groups	Any of pages, posts, videos, groups, people, places, events.
maxResultsPerType	integer	20	Unique results per keyword and result type (1–500).
maxItems	integer	1000	Hard cap on results for the whole run, across all keywords and types, also when keywords run in parallel.
sortPostsByRecent	boolean	false	Newest posts and videos first.
postsDateFrom	date	–	Drop dated posts and videos before this day (YYYY-MM-DD, a real calendar date). Native videos without a date are kept.
postsDateTo	date	–	Drop dated posts and videos after this day. Native videos without a date are kept.
requireKeywordMatch	boolean	true	Keep posts, videos and groups only when every word of the keyword appears in the visible post text, video title, author name or group name. See How keyword matching works.
proxyConfiguration	object	Apify datacenter proxy	Proxy for Facebook requests. The default works well.

Example input for daily brand monitoring:

{
    "searchQueries": ["nike air max", "#justdoit"],
    "resultTypes": ["posts", "videos", "groups"],
    "maxResultsPerType": 30,
    "sortPostsByRecent": true,
    "postsDateFrom": "2026-11-01"
}

Output example

A page result:

{
    "query": "nike",
    "rank": 10,
    "resultType": "pages",
    "facebookId": "100044519945902",
    "name": "Nike Football",
    "text": "Don't change your dream. Change the world. #justdoit",
    "url": "https://www.facebook.com/nikefootball",
    "isVerified": true,
    "websiteUrl": null,
    "searchModule": "ENTITY_PAGES",
    "scrapedAt": "2026-11-04T07:36:49.855Z"
}

A post result:

{
    "query": "nike",
    "rank": 1,
    "resultType": "posts",
    "facebookId": "1058803598947682",
    "text": "They’re ready to put the world on notice. 🗣️ From basement parties in the Bronx to the biggest stage in Paris this summer, breaking is re…",
    "url": "https://www.facebook.com/nike/posts/pfbid07zAJEC1jCvirZK6m5Rm8bqoBeK1chFnJhDJohdB4fZDPiCFyZgpRHSbet8eo9QgMl",
    "authorName": "Nike",
    "authorId": "100044541544829",
    "authorUrl": "https://www.facebook.com/nike",
    "groupId": null,
    "publishedAt": "2024-07-16T21:10:15.000Z",
    "reactionCount": 2181,
    "commentCount": 221,
    "shareCount": 241,
    "mediaType": "photo",
    "imageUrl": "https://scontent-sjc6-1.xx.fbcdn.net/v/t39.30808-6/...",
    "searchModule": "TOP_PUBLIC_POSTS",
    "scrapedAt": "2026-11-04T07:36:53.257Z"
}

A video result:

{
    "query": "nike",
    "rank": 1,
    "resultType": "videos",
    "facebookId": "1993703551290009",
    "text": "Nike Outlet clearance is GOOD right now. I stopped by my local Nike Factory Store and found some ser…",
    "url": "https://www.facebook.com/reel/1993703551290009/",
    "authorName": "Mama Deals",
    "authorUrl": "https://www.facebook.com/mamadeals1",
    "publishedAt": "2026-08-22T13:16:01.000Z",
    "reactionCount": 509,
    "commentCount": 14,
    "shareCount": 23,
    "viewCount": 31188,
    "durationSeconds": 26.8,
    "mediaType": "video",
    "searchModule": "VIDEOS_MBR_ENGAGEMENT"
}

A group result:

{
    "query": "nike",
    "rank": 2,
    "resultType": "groups",
    "facebookId": "1151067687031994",
    "name": "Nike Sneaker Exchange – Buy, Sell, Trade & Talk",
    "url": "https://www.facebook.com/groups/1151067687031994/",
    "groupId": "1151067687031994",
    "matchingPostCount": 1,
    "searchModule": "POSTS_IN_GROUP",
    "scrapedAt": "2026-11-04T07:37:06.212Z"
}

The example rows are shortened; real rows contain every field listed above, with null where a value does not apply.

How keyword matching works

With Only results that mention the keyword on (the default), each post, video and group row must match your keyword:

• The keyword is split into words at spaces and punctuation, so coffee shop chicago needs coffee, shop and chicago, and #blackfriday needs blackfriday.
• Matching ignores case and accents: café matches Cafe, and styled Unicode letters such as 𝐗 match x.
• Words of 3 or more characters match anywhere in the text, so nike also matches #nikerunning.
• Words of 1 or 2 characters, such as x, ai or f1, must appear as a whole word, so x does not match next.
• The words are checked in the post text, video title, author name and group name. Links are not checked.
• A keyword with no letters or digits at all (for example !!!) cannot be matched, and the run stops with a clear error before any charge. Turn the option off to search it anyway.

Pages, people, places and events are matched by Facebook itself on their names and are not filtered.

Why a run stops and how to read the result

The run stops collecting a keyword and type for one of these reasons, and the final status message says how many results were delivered out of the number requested:

• per-type limit reached: Max results per keyword and type was filled.
• Facebook had no more results: Facebook's result list ended.
• two pages in a row without a matching result: Facebook kept returning only filler content, so the Actor stopped early.
• page limit for this type reached: the Actor read its page limit for that type (about one page of 10 cards per 8 requested results, up to 60 pages).
• group scan limit reached: see the group note below.
• Max total results reached or maximum cost per run reached: the whole run stopped; later keywords may not be searched.
• keyword failed: Facebook kept failing for that keyword after retries; the other keywords still run.

The full breakdown per keyword and type is saved as RUN_SUMMARY in the run's key-value store, with delivered, requested, stopReason, failedKeywords and one line per keyword and type.

Tips and limits

• Groups come from matching posts. Facebook does not show its group search to logged-out visitors, so this is not Facebook's Groups tab. Groups are collected from the matching public posts that were shared in a group, during a bounded scan of post results. With only Groups selected, the scan covers about three times your per-type limit, at most 25 result pages per keyword. If you also select Posts, groups are collected throughout the longer post scan, which can run up to 60 pages per keyword at large per-type limits. Facebook shows a varying number of posts per page, so the number of posts scanned is approximate. The scan uses the same Most recent first and date settings as posts, so those settings change which groups you get. You often get fewer groups than the limit, and only groups with public posts. Use a specific topic keyword to find more.
• Logged-out results. Facebook shows logged-out visitors a smaller result set than a signed-in user sees, and its ranking differs. For a brand name, small look-alike pages can rank above the official page, so check isVerified and fetch a few more results than you need. Very deep searches stop when Facebook has no more results.
• Date windows are applied by Facebook at month level and then exactly by this Actor for every dated post and video. Native video cards that show no date cannot be placed in the window, so they are kept with publishedAt: null. Filter on publishedAt if you need dated rows only.
• Keyword filter. Turn off Only results that mention the keyword if you want everything Facebook returns for posts and videos, for example for misspellings or for posts that only show your brand in an image.
• Image links expire. Facebook image URLs are signed and stop working after some days. Download images soon if you need them.
• Failures. Each request is retried up to five times with a fresh anonymous session. If a keyword still fails, the run continues with the next keyword and reports the failed keyword in the log, the status message and the RUN_SUMMARY record. If charging or saving a result fails on the platform, the run fails with a message saying how many results were delivered, instead of saving uncharged rows.
• Public data only. Private profiles, private group posts and logged-in-only content are not returned.

Integrations and workflows

• Daily brand monitoring: schedule the Actor with sortPostsByRecent: true and yesterday's date as postsDateFrom, then send new rows to Slack or email with an Apify integration or webhook.
• Google Sheets dashboard: connect the dataset to Google Sheets and chart reactions and comments per keyword over time.
• Lead lists: search pages and places for <service> <city> keywords and push the rows with a websiteUrl into your CRM through Make, Zapier or n8n.
• Creator research: search videos for a niche topic, sort by viewCount, then enrich the top creators with Facebook Pages Scraper.
• Group discovery: collect groups for a topic, then read their posts with Facebook Group Posts Scraper.
• AI summaries: feed post text and engagement fields into an LLM step to summarise what people say about a campaign.

Use the Facebook Search Scraper with the Apify API

Run the Actor and get results with cURL:

curl -X POST "https://api.apify.com/v2/acts/automation-lab~facebook-multi-vertical-public-search-results/run-sync-get-dataset-items?token=YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"searchQueries":["nike"],"resultTypes":["pages","posts","videos","groups"],"maxResultsPerType":10}'

JavaScript (Node.js) with the apify-client package:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_API_TOKEN' });
const run = await client.actor('automation-lab/facebook-multi-vertical-public-search-results').call({
    searchQueries: ['nike', 'adidas'],
    resultTypes: ['pages', 'posts', 'videos', 'groups'],
    maxResultsPerType: 20,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.filter((item) => item.resultType === 'posts'));

Python with the apify-client package:

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")
run = client.actor("automation-lab/facebook-multi-vertical-public-search-results").call(run_input={
    "searchQueries": ["nike"],
    "resultTypes": ["posts", "videos"],
    "sortPostsByRecent": True,
    "maxResultsPerType": 25,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["resultType"], item["url"], item.get("reactionCount"))

Use with AI agents through MCP

You can use this Actor from Claude, Cursor and other MCP clients through the Apify MCP server.

Claude Code:

claude mcp add --transport http apify \
  "https://mcp.apify.com?tools=automation-lab/facebook-multi-vertical-public-search-results"

Claude Desktop, Cursor or VS Code (JSON configuration):

{
    "mcpServers": {
        "apify": {
            "url": "https://mcp.apify.com?tools=automation-lab/facebook-multi-vertical-public-search-results"
        }
    }
}

Selecting this Actor also adds the helper tools get-actor-run, get-dataset-items, get-key-value-store-record and abort-actor-run. A good agent workflow:

1. Start one run with call-actor and keep the returned run ID and dataset ID.
2. If the run is still running after the wait, check the same run with get-actor-run using short backoff (for example 2, 4, 8 seconds, up to 10) and a total deadline such as 120 seconds. Don't start a second run just to check progress.
3. If the run is still running at the deadline, stop polling and tell the user the run ID and that results are pending. Don't start another run; offer to check the same run later.
4. When it succeeds, read rows with get-dataset-items in pages using offset and limit: first offset=0, limit=20, then offset=20, offset=40 and so on, asking only for the fields you need (fields=resultType,url,text,reactionCount).
5. Keep two budgets while reading: at most 100 rows, and a total size budget for the rows you put in the conversation, for example 50 KB of serialized UTF-8 JSON. Use a smaller budget if your host or client has a lower context or tool-response limit. If a single page is already larger than the remaining budget, lower limit (down to 1 row) or drop long fields such as text, and skip a row that alone exceeds the budget.
6. Stop when either budget is used up or a page returns fewer rows than limit. Tell the user how many rows you read, the offset to continue from, and that the full result set is in the run's dataset (dataset ID, or download it as JSON, CSV or Excel from the run in Apify Console). Report partial results honestly if the run failed or you stopped early.
7. To explain missing results, read the RUN_SUMMARY record with get-key-value-store-record from the run's default key-value store.

Example prompts:

• "Search Facebook for 'oat milk' posts from the last week, most recent first, and summarise the top complaints."
• "Find 20 Facebook pages and places for 'yoga studio austin' and list the ones with a website."
• "Which Facebook groups are discussing 'tesla cybertruck'? Give me names, URLs and how many matching posts each had."

Is it legal to scrape Facebook search results?

This Actor collects only data that Facebook shows publicly to logged-out visitors. It does not log in, bypass privacy settings or access private groups or profiles. Web scraping of public data is generally lawful, but results can include personal data such as names of people who posted publicly. You are responsible for using the data lawfully, for example under GDPR or CCPA, and in line with Facebook's terms and your own purpose. Do not use the results for spam or harassment. If you're unsure, ask a lawyer. This Actor is not affiliated with or endorsed by Meta or Facebook.

When a run fails, the Actor sends sanitized diagnostic input, the error and the Actor, build and run IDs to our private GlitchTip error tracker so we can fix it. Secret fields and URL query strings are removed, and reports are kept for 30 days. The Actor does not use AI and does not send your input or results to an AI provider.

FAQ

Do I need a Facebook account or cookies? No. The Actor uses Facebook's public, logged-out search. You never provide credentials.

Why did I get fewer results than I asked for? Facebook returns fewer results to logged-out visitors, narrow keywords have fewer matches, and the keyword filter removes unrelated filler posts. Groups usually return fewer rows because they come only from a bounded scan of matching public posts. The status message and the RUN_SUMMARY record give the exact stop reason for every keyword and type.

Why are some posts old even with "Most recent first" off? By default Facebook ranks posts by relevance and popularity. Turn on Most recent first or set a date window for monitoring.

A run failed or returned nothing. What should I do? Check the log. "Rate limit exceeded" messages are retried automatically with new sessions. If every keyword fails, keep the default Apify proxy, or try the RESIDENTIAL proxy group, and run again. Nonsense or very rare keywords can return zero results without an error.

Can I search Facebook Marketplace or ads? No. Use Facebook Marketplace Public Listings Scraper or Facebook Ads Library Scraper.

Is the data real-time? Each run reads Facebook search when it runs. Schedule it to keep data fresh.

Related Facebook scrapers

• Facebook Pages Scraper – full details of pages you found in search.
• Facebook Pages Posts Scraper – the latest posts of specific pages.
• Facebook Group Posts Scraper – posts from public groups.
• Facebook People Search Scraper – deeper people directory search.
• Facebook Public Video Search Results – video-only search.
• Facebook URL to ID Resolver – convert Facebook URLs to numeric IDs.