Facebook Video Ads: HD Video Links from Ad Library

Search the Meta Ad Library for video ads only, by keyword, advertiser or country, and get each ad's MP4 video links (HD and SD) with its text, headline, button, landing page and start date.

Advertiser	Ad text	Headline	Button	Format	Started	Days active
Pureskin	Say goodbye to acne and hello to confidence 👋 ☁️ 1…	End Acne Forever.	Shop now	Video	2026-01-06	264
Beauty From Bees	If your skin feels tight, flaky, or itchy no matte…	20% OFF: Code SERUM20	Shop now	Video	2026-01-19	251
Eau Thermale Avène	Powerful cleanse, gentle on skin. NEW! Dermatologi…	Trusted by Dermatologists	Learn more	Video	2026-08-19	39

$0.60 per 1,000 ads on Starter ($0.75 Free, $0.50 Scale, $0.40 Business); with ad details $1.25 per 1,000 ($1.40 Free, $1.15 Scale, $1.05 Business). The prefilled run (up to 100 ads) costs about $0.06.

Use it for swipe files, creative research and hook libraries: long-running video ads (Days active) are usually the ones that work. No Facebook account, login or cookies.

Where the video links are

• videoUrl: the ad's video file (standard definition MP4), in the clean part of every row and in the 🎨 Creatives table.
• snapshot.videos[] (videoHdUrl, videoSdUrl, videoPreviewImageUrl): every video of the ad in HD and SD. Dynamic and carousel ads keep each card's media in snapshot.cards[], with the same videoHdUrl / videoSdUrl fields.
• image: the video's preview picture.

Download soon. The links are Meta's signed CDN links (fbcdn.net) and stop working about 5 days after the run. The Actor returns links, not the files.

What you can do with it

• Creative research / ad spy: every video ad running for "skincare", "running shoes" or any product, most impressions first, in any country.
• Swipe files and hook libraries: download the videos with their first lines, headlines and buttons.
• Competitor video ads: put a brand's Facebook page in Advertisers to pull just its video ads.
• New creatives as they launch: turn on Only new ads (under Filters) and schedule the run daily.

Input

Field	What it does
Search terms	Keywords, as in the Ad Library search box. One search per line.
Advertisers	A Facebook page link (facebook.com/nike), a page id (15087023444), an advertiser's Ad Library link, or the page name exactly as it appears on its ads.
Media type	Videos (the default). Pick all, images, memes, images and memes, or no media to get other ads instead.
Country	Two-letter code where the ads ran (US, GB, DE, ...) or ALL (US in the form, ALL if left out).
Max ads per search	Stop each search after this many ads (100 in the form, 200 if left out, up to 10,000). You pay per ad. Searches read about 1,000 ads in 3 minutes.
Include ad details	Adds the advertiser's profile and EU/UK reach data to each ad, at the with-details price (one more request per ad, so slower).

Under Filters and Ad Library URLs:

Field	What it does
Ad Library URLs	Searches copied from facebook.com/ads/library. The link's own filters are used.
Ad status	Active (default), inactive, or both.
Only new ads (monitor)	Return only ads started since the last run that this Actor hasn't returned before for the same search. You pay only for the new ads.
Ad category	All ads, or only ads about social issues, elections or politics (these carry spend, impressions and reach).
Exact phrase	Match the search terms as an exact phrase.
Shown on or after / before	The Ad Library's date filter: ads that were shown in that period, whatever day they started.

Give at least one search term, advertiser or Ad Library URL; a run with none fails with a message. Example:

{
  "searchQueries": ["skincare"],
  "advertisers": ["https://www.facebook.com/sephora/"],
  "mediaType": "video",
  "country": "US",
  "maxAdsPerQuery": 100
}

Output

One row per ad, in Meta's order (most impressions first; with Only new ads, most recent first). An ad found by several searches appears once, under the first search that found it.

Each row has two parts:

1. The clean fields first (below): flat, readable values for spreadsheets and the Console tables. Every row has them in the same order; a value Meta doesn't give is null.
2. Then the apify/facebook-ads-scraper fields, under that Actor's names: Meta's full ad record, with the HD links (below). They're what makes a row about 15 KB.

Clean fields, in row order:

• image: the creative's picture: the video's preview frame, its first image, or the first card's image.
• pageName: the advertiser (its Facebook page).
• adText, adTextSnippet: the ad's primary text in full, and its first 50 characters on one line.
• headline: the headline under the creative.
• adLibraryUrl: the ad in the Meta Ad Library.
• landingUrl: where the ad's button leads.
• ctaText: the button text, e.g. Shop now.
• format: Video, Dynamic creative, Carousel, Image, Catalog, ...
• platforms: Facebook, Instagram, Messenger, Threads, Audience Network, WhatsApp.
• isActive: still running when the run read it.
• pageLikes: likes of the advertiser's Facebook page.
• collationCount: how many ads use this creative and text. Meta says it on one ad of each group, so it's often null.
• daysActive: days from startedAt to lastShownAt.
• startedAt: the day the ad started (YYYY-MM-DD).
• lastShownAt: the last day it ran: the day it stopped, or the run's day for active ads.
• rank: the ad's place in its search (1 = most impressions).
• videoUrl: the ad's video file (standard definition), for video ads.
• pageUrl: the advertiser's Facebook page.
• adArchiveId, pageId: the ad's Library ID and the advertiser's page id.
• searchInput: the search term, advertiser or Ad Library URL that found the ad.
• scrapedAt: when the run started (UTC).

Dynamic and catalog ads keep placeholders like {{product.name}} in their top text; the clean fields take the text, headline and picture of the ad's first card instead, as the Ad Library shows it.

When a search is for Issues, elections or politics (Ad category), the clean part also has spend, spendMin, spendMax, currency, impressions, reachEstimate and paidFor after isActive; Meta publishes them for those ads only. With Include ad details, detailsStatus, isPageVerified, instagramUrl, instagramFollowers, euReach, ukReach, targetAges, targetGender, reachByCountry, payer and beneficiary come in after pageUrl (an ad whose details are missing is charged as a plain ad).

In the Console, the dataset has these tables: 📊 Overview (creative, advertiser, text, headline, links, button, format, platforms, active, started, days active, search), 📈 Stats (rank, days active, page likes, same-creative count, dates), 🎨 Creatives (picture, text, headline, button, video, landing page), 🏛️ Political (fills on political searches) and 📋 Details (fills with Include ad details). All fields shows the whole row, the facebook-ads-scraper part included.

facebook-ads-scraper fields

After scrapedAt, every row has the same fields as apify/facebook-ads-scraper, under the same names: Meta's raw ad record with camelCase keys, plus inputUrl, pageID / adArchiveID and startDateFormatted / endDateFormatted. Pipelines built on that Actor keep working when you switch. This part is nested (snapshot holds the creative), uses Meta's constants (FACEBOOK, VIDEO), and has every image and video variant (HD and SD).

• snapshot: the creative. It holds body.text, title, caption, ctaText, linkUrl, displayFormat (IMAGE, VIDEO, CAROUSEL, DCO, DPA), images, videos (each with videoHdUrl, videoSdUrl, videoPreviewImageUrl), cards (carousel and dynamic-ad variants, each with its own text and media), and the page's likes, categories and picture.
• adArchiveID, pageID: the ad's Library ID and the advertiser's page id (the same as adArchiveId and pageId).
• startDateFormatted / endDateFormatted (also startDate / endDate in unix seconds), publisherPlatform.
• spend, impressionsWithIndex, reachEstimate, currency: Meta fills these only for political and issue ads.

Run summary

The run's OUTPUT record lists every search with Meta's result count (metaCount), the ads read, the ads delivered and why it stopped: complete, maxAds, chargeLimit, caughtUp, readCap, duplicate, or a failure: blocked, error, incomplete.

Reliability

• Uses Apify residential proxies by default. Meta refuses most datacenter IPs after the first page of results, so datacenter proxies aren't recommended.
• A refused or failed request is retried on a new IP.
• If Meta still refuses a search, the ads already collected are kept and the run fails with the reason. It doesn't report success on an incomplete search.
• Each search is checked against the result count Meta reports. A search whose results end well short of that count fails the run.
• Each ad appears once per run, even when several searches find it.

Pricing

Pay per ad, by your Apify plan. No start fee, no monthly fee.

Per 1,000 ads	Free	Starter	Scale	Business
Ad	$0.75	$0.60	$0.50	$0.40
Ad with details	$1.40	$1.25	$1.15	$1.05

• The with-details price is the ad price plus a flat $0.65 per 1,000 for the details, the same on every plan.
• An ad is charged the with-details price only when its details came; an ad whose details failed is charged as a plain ad.
• Set a maximum cost per run in the run options: the run stops before it would go past it and keeps what it collected (OUTPUT.chargeLimitReached is then true).

FAQ

Do I need a Facebook account? No. Everything comes from the public Ad Library, without logging in.

Does it download the video files? No: it returns the MP4 links, which you can download with any tool or script. Download them within a few days; Meta's links expire.

Why is a row's format "Dynamic creative"? Meta shows these ads in several versions; the videos are in snapshot.cards[].

Which countries work? Every country in the Ad Library, or ALL.

Feedback

Found a bug or missing a field? Open an issue on the Issues tab and we'll look at it within 48 hours. Happy with the data? A short review on the Store helps others find this Actor.