Google Ads Transparency Scraper — Advertisers, Creatives & Landing Domains

Pull every ad a company is running through Google Ads, Search, Display, YouTube and Shopping — by website domain, advertiser name or advertiser ID. No headless browser, no login, no cookies to paste.

---

The failure mode that breaks most Ads Transparency scrapers

The Ads Transparency Center has no public API. Everything goes through an internal RPC endpoint that answers happily for a while and then stops.

Measured on 2026-09-20: after roughly 30 requests from a single IP, the endpoint stops returning JSON and starts returning a 3.3 KB HTML rate-limit page — sometimes with a 429, sometimes not. A scraper that only checks the status code, or that assumes the body is JSON, will either crash or hand you an empty result and call it a win.

This Actor treats every one of those cases as a block:

• Non-JSON body on a 200 response → blocked.
• 429, 403, 503 → blocked.
• Socket-level hang → blocked. A hard AbortSignal cuts the request instead of letting a dead proxy tunnel burn two minutes of compute.

On a block it throws the whole session away — IP and cookies together — takes a new proxy session, backs off exponentially and retries. Refreshing cookies alone does nothing once the IP is on Google's list.

What else is different

• No headless browser. Data comes straight from the JSON RPC. A 32-ad run finishes in about two minutes on 1 GB of RAM.
• Every entry type in one field. Domain, advertiser name, advertiser ID or a full Ads Transparency Center URL — the type is detected per entry, so you can mix them freely.
• Multi-account brands are merged. Big advertisers run ads from several accounts (Nike alone shows up under Nike, Inc., NIKE GLOBAL TRADING B.V. and its media agency). Searching by name pulls the top matching accounts together instead of silently returning one of them.
• Per-country last-shown dates. With creative details on, each ad reports every country it ran in and when it was last seen there — resolved to real ISO country codes, all 249 of them.
• Partial success. A query that resolves to nothing returns a row with error and errorType, not a dead run.

Input

Field	Type	Notes
queries	array	Domain (nike.com), advertiser name (Coca-Cola), advertiser ID (AR1673…) or an Ads Transparency Center URL. Mix freely
region	string	ISO country code, or anywhere for worldwide
maxAdsPerQuery	integer	Stop after this many ads per entry. Default 50
maxAdvertisersPerQuery	integer	How many advertiser accounts to merge when the entry is a name. Default 2
includeCreativeDetails	boolean	Adds every creative variant and the per-country last-shown dates. One extra request per ad
requestDelayMs	integer	Default 700. Lower is faster but hits the rate limit sooner
proxyConfiguration	object	Strongly recommended. Residential proxies give the highest success rate

Example

{
  "queries": [
    "nike.com",
    "Coca-Cola",
    "AR16735076323512287233",
    "https://adstransparency.google.com/advertiser/AR16735076323512287233?region=US"
  ],
  "region": "US",
  "maxAdsPerQuery": 50,
  "includeCreativeDetails": true
}

Output

One row per ad creative.

{
  "advertiserName": "Nike, Inc.",
  "advertiserId": "AR16735076323512287233",
  "advertiserLocation": "US",
  "creativeId": "CR03886179215074131969",
  "adUrl": "https://adstransparency.google.com/advertiser/AR16735076323512287233/creative/CR03886179215074131969?region=US",
  "creativeKind": "image",
  "formatCode": 1,
  "imageUrl": "https://tpc.googlesyndication.com/archive/simgad/2393121755386194318",
  "imageWidth": 348,
  "imageHeight": 219,
  "previewUrl": null,
  "videoId": null,
  "landingDomain": "nike.com",
  "firstShownAt": "2023-11-16T22:13:08.000Z",
  "lastShownAt": "2026-09-20T04:12:13.000Z",
  "variantCount": 5,
  "shownInRegions": [{ "region": "US", "lastShownDate": "2026-09-01" }],
  "query": "nike.com",
  "queryType": "domain",
  "region": "US",
  "error": null,
  "errorType": null,
  "scrapedAt": "2026-09-20T04:29:40.713Z"
}

creativeKind is derived from the creative payload itself: image when the ad ships a static image, rendered when Google returns a rendered preview frame, video when a YouTube video is attached. formatCode is the raw internal format value, exposed as-is rather than guessed at.

Failed queries

{
  "query": "not-a-real-domain-xyz123.com",
  "queryType": "domain",
  "error": "No ads found for \"not-a-real-domain-xyz123.com\" in region US.",
  "errorType": "NO_RESULTS"
}

errorType is one of NO_RESULTS, NOT_FOUND, BLOCKED, REQUEST_FAILED, DETAIL_UNAVAILABLE, INVALID_INPUT, UNEXPECTED_ERROR.

Run summary

Every run writes RUN_SUMMARY to the key-value store, so an automation can assert on it instead of counting rows:

{
  "requested": 5,
  "succeeded": 4,
  "failed": 1,
  "successRate": 80,
  "adsCollected": 32,
  "retries": 0,
  "detailFailures": 0,
  "proxySessionsUsed": 2,
  "region": "US"
}

What people use it for

• Competitor ad monitoring. Which creatives a rival is running right now, and which ones they quietly killed.
• Creative research. Pull hundreds of live ads in a vertical before briefing a new campaign.
• Agency and affiliate discovery. landingDomain reveals which media agencies and affiliates buy traffic for a brand.
• Ad longevity signals. firstShownAt and lastShownAt show which creatives survived months of optimisation — the ones worth copying.
• Political and regulated-ad tracking. Region filtering works on every country Google reports.

Notes and limits

• Only ads Google publishes in the Ads Transparency Center are available. Google caps very large advertisers, so a brand listed as "~9,000 ads" will not return all of them.
• Ad counts shown in the Transparency Center are approximate ranges by design.
• Searching by advertiser name needs an exact-ish match. If a name returns nothing, use the website domain instead — it is the more reliable key.
• includeCreativeDetails doubles the request count. Leave it off for broad sweeps, turn it on when you need variants and per-country dates.
• Every returned row, including rows that carry an error, counts as one result.

Found a bug or need a field that is missing? Open an issue on the Actor page.

Code examples

Copy-paste Python and Node examples (zero dependencies) live in github.com/steadyapi/scrapers. The Python one for this Actor is python/google_ads_transparency.py.

Minimal call with plain HTTP:

curl -X POST "https://api.apify.com/v2/acts/steadyapi~google-ads-transparency-scraper/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H 'Content-Type: application/json' -d @input.json

Use it from AI agents (MCP)

Add this Actor as a tool in Claude, Cursor or any MCP client through the Apify MCP server:

https://mcp.apify.com/?tools=steadyapi/google-ads-transparency-scraper

Authenticate with your Apify token (OAuth is also supported). The agent can then call the Actor directly and read the results.

FAQ

What can I search by? A website domain (nike.com), an advertiser name (Coca-Cola), an advertiser ID (AR...) or a full Ads Transparency Center URL. Mix them in one run.

Does it include YouTube and display ads? Yes. Search, display and video creatives are returned, each with its format, first and last shown dates and preview links.

Can I limit results to one country? Set region to a country code, or anywhere for worldwide results with the list of regions each ad ran in.

Do I need a Google Ads account? No. The Ads Transparency Center is public and no login is used.