Google Ads Transparency & Ad Library Scraper

Scrape the Google Ads Transparency Center (Google's ad library) for any domain, advertiser ID or advertiser name and get every ad with days running, first/last shown dates, the countries it ran in and the full text-ad copy (headline, description, display URL) — including responsive search ads and ads Google only stores as screenshots.

Why the screenshots matter: for many big advertisers, Google keeps the text ads only as images — in our tests, 149 of 150 booking.com and 120 of 120 expedia.com text ads had no text version at all. This Actor reads those screenshots, so you still get the headline and description.

Find an advertiser's long-running winners in one run: filter by minimum days running and sort by longest running.

What makes it different

	This Actor	Typical Ads Transparency scrapers
Domains, advertiser IDs, advertiser names, Transparency Center links	✅ any mix	⚠️ some
Minimum days running filter (find ads that work)	✅	❌
Sort by longest running / newest	✅	❌
Days shown, first & last shown date, active now	✅	⚠️
Countries per ad with last shown date per country	✅ included	⚠️ extra charge
Text-ad copy from Google's text preview	✅	⚠️
Responsive search ad copy (headline, description, business name)	✅	❌
Copy from screenshot-only text ads (OCR, 10+ scripts)	✅	⚠️ some
Country, platform (Search, YouTube, Shopping, Maps, Play), format and date filters	✅	⚠️
Only new ads since the last run (for monitoring)	✅	⚠️
Failed rows are free	✅	❌

Use cases

• Competitor ad research — see every ad a competitor runs, where, and for how long.
• Find winning ads — minDaysRunning: 30 + sortBy: longestRunning shows the ads an advertiser keeps paying for.
• Ad copy swipe files — collect headlines and descriptions across a market, in any language.
• Market entry & geo research — which countries a brand advertises in, and since when.
• Monitoring — schedule daily runs with onlyNewAds to get only fresh creatives.
• Agencies & SEO/PPC tools — feed competitor ad data into dashboards, Sheets or your own app.

Input

Field	Description	Default
domains	Advertiser websites, e.g. nike.com (subdomains included). Transparency Center links with ?domain= also work	—
advertiserIds	Advertiser IDs (AR…) or adstransparency.google.com/advertiser/AR… links	—
queries	Advertiser names, e.g. Samsung Electronics — looked up on the Transparency Center	—
region	Country where the ads were shown (2-letter code) or anywhere	anywhere
platform	all, search, youtube, shopping, maps, play	all
format	all, text, image, video	all
dateFrom / dateTo	Only ads shown in this period (YYYY-MM-DD)	—
minDaysRunning	Only ads shown on at least this many days	0
sortBy	recentlyShown (Google's order), longestRunning, newest	recentlyShown
onlyNewAds	Return only ads not returned to you in earlier runs (same domain/advertiser)	false
maxAdsPerTarget	Max ads per domain / advertiser / name	200
maxResults	Hard cap on ads across all targets (0 = no limit)	0
includeRegions	Add countries per ad (same price)	true
includeAdCopy	Extract text-ad copy (charged as ad with copy when found)	true
advertisersPerQuery	Matching advertisers scraped per name	1
maxScanPerTarget	Ads scanned per target before filtering/sorting (scanning is free)	1000

You can put any input in any list — a domain typed into queries or an AR… ID typed into domains is recognized automatically.

Example: a competitor's long-running search ads in the US

{
  "domains": ["booking.com"],
  "region": "US",
  "platform": "search",
  "format": "text",
  "minDaysRunning": 30,
  "sortBy": "longestRunning",
  "maxAdsPerTarget": 100
}

Output

One row per ad:

{
  "input": "nike.com",
  "inputType": "domain",
  "advertiserName": "Nike, Inc.",
  "advertiserId": "AR16735076323512287233",
  "domain": "nike.com",
  "creativeId": "CR13173014960004923393",
  "format": "text",
  "firstShown": "2022-11-30",
  "lastShown": "2026-10-30",
  "daysShown": 1383,
  "isActive": true,
  "headline": "Nike® Official Site",
  "description": "Explore new Nike® sportswear and find your next everyday essential.",
  "displayUrl": "nike.com",
  "businessName": "Nike",
  "copySource": "text",
  "copyConfidence": null,
  "imageUrl": "https://tpc.googlesyndication.com/archive/simgad/...",
  "previewUrl": "https://displayads-formats.googleusercontent.com/ads/preview/content.js?...",
  "regions": [
    { "country": "US", "countryName": "United States", "lastShown": "2026-10-30" }
  ],
  "regionCount": 1,
  "variations": 2,
  "transparencyUrl": "https://adstransparency.google.com/advertiser/AR16735076323512287233/creative/CR13173014960004923393?region=anywhere",
  "detailStatus": "ok",
  "copyStatus": "ok",
  "scrapedAt": "2026-10-30T12:44:01+00:00"
}

• daysShown — number of days the ad was shown (from Google). isActive — last shown today or yesterday (UTC).
• regions — every country the ad ran in, with the last date shown there.
• copySource — text (read from Google's text preview, exact) or ocr (read from the ad screenshot; copyConfidence 0–100).
• copyStatus — ok, noCopy (no text available, e.g. image-only template), lowConfidence (OCR confidence below 90 — text kept but may contain errors, charged as a plain ad), failed, or null for image/video ads.
• headline is null for Dynamic Search Ads, where Google generates the headline from the landing page.
• {Keyword:…} in a headline is the advertiser's keyword-insertion placeholder, kept as written.

Pricing

Pay per event — you pay only for the ads you get, no platform usage fees:

• Ad — $2.00 per 1,000 (all fields above, countries included)
• Ad with copy — $3.50 per 1,000 (text ads where the headline/description was extracted — charged instead of the plain ad)

Image and video ads, and text ads without extractable copy, are plain ads ($2.00 per 1,000). Rows where the ad details could not be loaded (detailStatus: "failed") are free.

Examples:

• 1,000 ads, mixed formats, 40% text ads with copy → 400 × $0.0035 + 600 × $0.002 = $2.60
• 1,000 image/video ads → $2.00
• Daily monitoring of 10 competitors with onlyNewAds (≈ 20 new ads/day) → ≈ $0.05 per day

Set a maximum cost per run in the run options — the Actor stops before exceeding it and does not request ads it cannot charge for. The exact number of charged events is saved in the run's key-value store as RUN_STATS.

Reliability

Requests are spread over residential proxies where Google rate-limits, retried with a fresh IP when Google answers with "too many requests", and the run stops by itself instead of hammering Google if blocking becomes frequent — you keep (and pay only for) the ads already saved. CAPTCHAs are never solved or bypassed.

In our tests on large advertisers (booking.com, expedia.com, samsung.com, hubspot.com, nike.com):

• 602 of 603 ads loaded with full details (countries, variations) — the one failed row was free.
• High-confidence copy for 127 of 150 text ads (85%) in a Nov 1, 2026 test (30 text ads each) — 29 from Google's text preview, 98 by OCR of screenshot-only ads. Another 19 got OCR text marked lowConfidence (returned, but charged as a plain ad) and 2 had no copy. Only the high-confidence ads are charged as ad with copy.

Monitoring with Schedules

Use Schedules in Apify Console with onlyNewAds: true to get only new creatives every day. Results can be downloaded as JSON, CSV or Excel, or sent to Google Sheets, Slack, webhooks and other apps via Apify integrations.

Good to know

• A domain search returns what Google returns for that domain on the Transparency Center: ads that link to the domain or its subdomains — sometimes from other advertisers (e.g. landing pages hosted on a platform's subdomains). Use advertiserIds for one exact advertiser, or filter by advertiserName.
• Data comes from Google's public Ads Transparency Center. Ad counts shown there are ranges (e.g. "50,000+"); this Actor returns the ads themselves.
• Many big advertisers' text ads are stored by Google only as screenshots — the Actor reads them with OCR. OCR text can contain small errors (e.g. "Al" vs "AI", currency symbols) and words cut off where Google's screenshot crops the ad; check copyConfidence and copySource.
• firstShown can't go back further than Google's own records: in our tests the oldest ads all showed 2021-10-25, which looks like where the Transparency Center's history starts. Read that date (and the daysShown it implies) as "running since at least then".
• OCR is the slowest step: expect ~50 text ads with copy per minute (image/video ads are much faster). Turn off includeAdCopy if you don't need copy.
• Shopping ad product titles and political ad spend are not included yet.
• Only public ad information about advertisers (companies) is collected. No personal data.

FAQ

Can't I just use Google's free BigQuery dataset? It covers only ads shown in the EEA and Turkey, has no ad copy or images, and can lag behind the website. This Actor works worldwide, returns headlines and descriptions, and reads the live site.

Can I get all ads of a very large advertiser? Yes — raise maxAdsPerTarget (up to 10,000 per target). For very large advertisers, narrow with region, platform, format or dates.

How do I find winning ads? Set minDaysRunning (e.g. 30 or 90) and sortBy: longestRunning. Ads that run for months are usually the ones that convert.

Why is the headline empty for some ads? Dynamic Search Ads have no fixed headline (Google writes it from the landing page), and some ad templates are image-only.

Which languages does ad-copy OCR support? Latin-script languages (English, Spanish, French, German, Turkish, Vietnamese…), Korean, Japanese, Simplified & Traditional Chinese, Hebrew, Cyrillic, Greek, Arabic and Thai. The script is chosen from the advertiser name and the countries the ad ran in.

What does onlyNewAds remember? The creative IDs returned to you before for the same domain/advertiser, stored in your own Apify account. The first run returns all ads.

Tutorial

How to Export Google Ads Transparency Center Data (with Ad Copy) in Python — why most text ads exist only as screenshots, how to OCR them yourself, and a worked example finding a competitor's long-running ads.

---

This Actor is not affiliated with, endorsed by or sponsored by Google. It only collects publicly available ad information from the Google Ads Transparency Center.