Facebook Ad Library Full Scraper

Pull complete ad records out of Meta's public Ad Library — the full ad copy, the headline, the call-to-action, the destination URL, every image and video file including carousel cards, the flight dates, how many days the ad has been on air, which Meta surfaces it runs on, and who is paying for it.

Search by keyword (everything a niche is running) or by advertiser (everything one page is running), in any country the Ad Library covers.

No login, no cookies to paste, no proxy required.

What you get

One row per ad:

Field	Example
pageName, pageId, pageUrl	Nordstrom Rack, 89516513179, …
adText	Buy online and pick up in store for free! Get up to 70% off …
title, linkDescription, caption	Fast & Free Pickup, …, nordstromrack.com
ctaText, ctaType	Learn more, LEARN_MORE
linkUrl	where the click goes
displayFormat	VIDEO, IMAGE, DCO, …
imageUrls, videoUrls, thumbnailUrls	direct CDN files
cards[]	per-card copy for carousel ads
startDate, endDate, runningDays, isActive	2026-09-10T07:00:00Z, …, 9, true
publisherPlatforms	["FACEBOOK","INSTAGRAM","MESSENGER"]
pageLikeCount, pageCategories	2518041, ["Clothing (Brand)"]
adLibraryUrl	open the same ad in a browser
totalMatchingAds	how many ads the search matches in total

Plus a RUN_SUMMARY row with ads collected, how many are active, how many distinct advertisers and the median days on air.

Two things worth knowing before you buy

1. Spend and impressions exist only for political and issue ads. spendText, impressionsText and targetedCountries are populated when adType is political_and_issue_ads. For ordinary commercial ads Meta does not publish them anywhere, at any price, so they come back empty. That is Meta's policy, not a gap in this actor — and no scraper can fill it. Everything else in the table above is available for commercial ads.

2. Demographic targeting is not published for commercial ads either. The "target wilayah/demografi" you may be looking for is available as targetedCountries on political and issue ads. Age/gender breakdowns are shown in Meta's own EU transparency reports for a subset of ads and are not part of the Ad Library search response. This actor returns what the Ad Library actually serves and leaves the rest null rather than inventing a plausible number.

Advertiser input takes a numeric page id

The Ad Library addresses advertisers by numeric page id, not by vanity slug. Open the advertiser in the Ad Library and copy the URL — it contains view_all_page_id=<id> — or paste the id on its own. A plain facebook.com/<slug> URL is rejected with a log line rather than silently returning nothing.

Advertiser searches include the influencer ads a brand pays for

Search one advertiser and a few rows will carry somebody else's name in pageName — a creator. Those are branded-content ads: the creator's page posts them, the brand pays. The row tells you both: isBrandedContent: true, sponsorPageName: "Nordstrom Rack". Nothing has leaked; this is part of what that advertiser is running, and it is usually the part a competitor analysis wants to see.

Input example

{
  "searchTerms": ["running shoes", "protein powder"],
  "countries": ["US", "GB"],
  "adType": "all",
  "activeStatus": "active",
  "mediaType": "video",
  "maxAdsPerQuery": 120,
  "minRunningDays": 14,
  "exportFormats": ["csv"]
}

That runs four searches (2 terms × 2 countries), keeps video creatives that have been on air at least two weeks, and drops a CSV in the key-value store alongside the dataset.

How many ads can it get?

Facebook server-renders the first 30 ads of a search. Beyond that, this actor replays Facebook's own pagination query, 30 ads at a time, up to maxAdsPerQuery. A search matching 26,000 ads will happily give you 1,000 of them; set the cap to what you actually want to pay for.

If Facebook reshuffles its JavaScript bundles mid-run and the pagination query cannot be located, the actor keeps the 30 server-rendered ads, logs a warning and carries on. You get a smaller honest answer, never a failed run.

Common uses

• Competitor monitoring — track every ad a competitor is running, and how long each has survived. A creative on air for 90 days is one that works.
• Swipe files — collect the copy and creative of a whole niche in one run.
• Landing-page research — linkUrl shows the exact destination of every ad.
• Political transparency — spend and impression ranges, per ad, per country.

Notes

• Dynamic-creative ads legitimately contain Meta template tokens such as {{product.name}} in adText. That is the published text, not a parse error.
• videoUrls and imageUrls are signed CDN links and expire after a while. Download what you need during or shortly after the run.
• endDate is filled in for running ads too — it is the current end of the flight, not a stop date. Read it with isActive.
• The Ad Library answers the first request of a session with HTTP 403 and a verification challenge. This actor solves it automatically; nothing is required from you.