Google Ads Transparency Scraper avatar

Google Ads Transparency Scraper

Pricing

from $1.50 / 1,000 ads

Go to Apify Store
Google Ads Transparency Scraper

Google Ads Transparency Scraper

Export the ads any advertiser runs on Google (Search, YouTube, Display, Shopping, Maps) from the Ads Transparency Center by name, domain or ID, with region, date and format filters, ad details, and a monitor mode that returns only new ads on scheduled runs.

Pricing

from $1.50 / 1,000 ads

Rating

0.0

(0)

Developer

Dong Nguyen

Dong Nguyen

Maintained by Community

Actor stats

0

Bookmarked

2

Total users

1

Monthly active users

9 hours ago

Last modified

Features

Limited permissions

Share

Export the ads any company runs on Google — Search, YouTube, Display, Shopping, Maps — from Google's public Ads Transparency Center. Give it advertiser names, domains, advertiser IDs or Transparency Center URLs and get one clean row per ad: who ran it, the format, when it first and last ran, where, and a link to the creative.

No login, no cookies, no browser. The Actor reads the same public data the Transparency Center website shows.

Who it is for

  • Marketers and agencies — see every ad a competitor is running right now, how long each has been live, and which creatives they keep.
  • Competitive-intelligence teams — track when a brand enters a country, which formats it favours, and how many creatives it rotates.
  • Weekly competitor watch — turn on monitor mode, schedule the task, and each run returns only the creatives launched since the last one.
  • AI-agent builders — a predictable JSON shape (every item has the same keys, null when unknown) that is safe to feed to a model or a pipeline.

What you get per ad

FieldMeaningNeeds includeDetails
advertiserId, advertiserNameGoogle's advertiser ID (AR…) and display nameno
creativeIdGoogle's ad ID (CR…)no
formattext, image or videono
firstShown, lastShownFirst and last time Google served the ad (UTC)no
daysShownNumber of days the ad was servedno
landingDomainDomain the ad points to. Always filled for domain queries; for other queries only when it can be read from the creativeno
previewUrlThe creative: an image URL when Google stores a static rendering, otherwise Google's preview script URLno
imageUrlStatic rendering of the ad, when one exists (most text ads, some image ads)no
transparencyUrlLink to the ad on the Transparency Centerno
regionsCountries (ISO codes) where the ad was shown. Without details: the region you filtered by, or null for "anywhere"full list: yes
regionStatsPer country: first/last shown date, and impression range where Google publishes ityes
impressionsMin, impressionsMaxTotal impression range. Google only publishes this for ads shown in the EU and similar regulated regions; elsewhere nullyes
variationCount, assetUrlsNumber of creative variations and the image URL of each; video thumbnails for video adsyes
videoUrlYouTube link for video adsyes
adText, adTextLines, displayUrlVisible ad copy. Only for text ads that Google serves as HTML — see limits belowyes
advertiserLegalName, advertiserCountry, advertiserVerifiedAdvertiser identity as shown by Googleyes
query, queryType, detailsFetched, scrapedAtWhich input produced the row, how it was interpreted, whether details were loaded, and whenno

Input

FieldTypeDefaultNotes
queriesarray of strings— (required)Domain, advertiser name, advertiser ID or Transparency Center URL. Mix freely.
regionstringanywhereISO country code (US, GB, DE, VN …) or anywhere.
dateFrom, dateToYYYY-MM-DDnoneOnly ads shown in this window.
formatall / text / image / videoall
platformall / search / youtube / maps / play / shoppingallOnly ads shown on that Google platform — the Transparency Center's own Platform filter. Google applies it to ads shown since 4 Sep 2023 only.
maxAdsPerQueryinteger1000 = no limit.
includeDetailsbooleanfalseAdds the detail fields above. One extra request and one extra charge per ad; much slower.
onlyNewAdsbooleanfalseMonitor mode: return and charge only ads this monitor has not returned before and that were first shown since its last run. See below.
monitorNamestringdefaultSeparate monitors keep separate memories, e.g. brand-a-us.
maxAdvertisersPerNameinteger1For name queries: how many matching advertisers to include.
maxRequestsPerSecondnumberautoAuto = 1 when listing, 0.25 with details.
proxyConfigurationApify proxy inputnoneOptional. See "Rate limits" below.

How each kind of query is handled

  • Domain (nike.com) — ads from every advertiser that points ads at that domain (often several legal entities, resellers and affiliates). This is usually what you want for competitor research.
  • Advertiser ID (AR16735076323512287233) — exactly that advertiser. The most precise option.
  • Advertiser name (Nike) — resolved through Google's own suggestions to the advertiser whose name matches and who has the most ads (here: "Nike, Inc."). Names are ambiguous; the log and the advertiserId field tell you who was picked. Raise maxAdvertisersPerName or use the domain if you want them all.
  • Transparency Center URL — an advertiser page, a single-ad page (/advertiser/AR…/creative/CR…, returns that one ad with details), or a ?domain= search. Filters in the URL are ignored; the input fields apply.

Examples

Everything pointing at a competitor's domain in the US, up to 500 ads:

{
"queries": ["nike.com"],
"region": "US",
"maxAdsPerQuery": 500
}

Video ads of three brands in Germany during Q3, with details:

{
"queries": ["zalando.de", "AR01625195283841286145", "HubSpot"],
"region": "DE",
"format": "video",
"dateFrom": "2026-07-01",
"dateTo": "2026-09-30",
"maxAdsPerQuery": 50,
"includeDetails": true
}

One specific ad:

{
"queries": ["https://adstransparency.google.com/advertiser/AR15589870916631265281/creative/CR16333246845520183297?region=DE"]
}

Output sample

A real item from a run on 19 November 2026 ("queries": ["shopify.com"], "includeDetails": true). regionStats is shortened here from 13 entries to 2; nothing else is edited.

{
"query": "shopify.com",
"queryType": "domain",
"advertiserId": "AR01625195283841286145",
"advertiserName": "Shopify Inc.",
"creativeId": "CR17723666237570940929",
"format": "video",
"firstShown": "2026-10-06T16:36:23Z",
"lastShown": "2026-11-18T12:01:58Z",
"daysShown": 44,
"regions": ["DZ", "ZA", "EG", "LY", "NG", "ZW", "MA", "TZ", "KE", "UG", "TN", "GH", "SN"],
"landingDomain": "shopify.com",
"previewUrl": "https://displayads-formats.googleusercontent.com/ads/preview/content.js?client=ads-integrity-transparency&obfuscatedCustomerId=6178574835&creativeId=827208976688&uiFeatures=12,54&adGroupId=192321953246&assets=%3DH4sIAAAAAAAAAONy4uLkWPfiV8duHgEeLk6Olis9t3bxCDBzcXJ0QZiMXJwc7Qjm-51PPj5iF9Dk4uTovLRr8zwuATYAu27RUkQAAAA&sig=ACiVB_xhYFwoyhfNoDuB8wUxtmgHUg0HCQ&htmlParentId=fletch-render-1317732629125866294&responseCallback=fletchCallback1317732629125866294",
"imageUrl": null,
"videoUrl": "https://www.youtube.com/watch?v=acYbMCenDg8",
"assetUrls": ["https://i.ytimg.com/vi/acYbMCenDg8/hqdefault.jpg"],
"adText": null,
"adTextLines": null,
"displayUrl": null,
"variationCount": 6,
"regionStats": [
{ "region": "DZ", "firstShown": null, "lastShown": "2026-10-26", "impressionsMin": null, "impressionsMax": null },
{ "region": "EG", "firstShown": null, "lastShown": "2026-11-19", "impressionsMin": null, "impressionsMax": null }
],
"impressionsMin": null,
"impressionsMax": null,
"advertiserCountry": "CA",
"advertiserLegalName": "Shopify Inc.",
"advertiserVerified": true,
"transparencyUrl": "https://adstransparency.google.com/advertiser/AR01625195283841286145/creative/CR17723666237570940929?region=anywhere",
"detailsFetched": true,
"scrapedAt": "2026-11-19T13:01:43Z"
}

When a query cannot produce ads, the dataset gets a notice row instead of the run failing. Notice rows are never charged. Their shape:

{
"query": "zzqxqxnotexistingadvertiser",
"queryType": "name",
"errorType": "not_found",
"error": "No advertiser matches 'zzqxqxnotexistingadvertiser'. Try the advertiser's domain or its AR… id.",
"scrapedAt": "2026-11-19T13:30:00Z"
}

errorType is one of invalid_query, not_found, no_results, request_failed, blocked, internal_error.

Monitor mode: only new ads

Set onlyNewAds: true and schedule the run (daily or weekly). The first run returns the current ads as usual and saves them as the baseline. Every later run returns only ads that

  1. this monitor has not returned before, and
  2. Google first showed on or after the previous run (minus a 2-day margin, because Google dates ads by day and publishes them with a delay).

Older ads and ads you already have are skipped before any detail lookup, so they are never charged. Google lists ads by the time they were last shown, newest first, so the Actor stops paging an advertiser once a whole page was last shown before the previous run.

The memory lives in your own named key-value store google-ads-transparency-monitor, one record per monitorName (the ad IDs plus the time of the last run). Delete the record to start a monitor over. Use one monitorName per task so two schedules never share a memory.

Tested live on 21 Nov 2026, shopify.com, region US, maxAdsPerQuery: 20, run twice a few minutes apart: run 1 returned 20 ads (the baseline); run 2 read 11 list pages without charge, skipped 20 known and 1,017 older ads, and returned 20 ads all first shown on 19–20 Nov 2026 — none of them from run 1.

Use it from an AI agent (MCP)

Apify's hosted MCP server exposes this Actor as a tool, so Claude, Cursor, VS Code or any other MCP client can call it directly. Server URL:

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

Authenticate with your Apify API token as a bearer header. In Claude Code, for example:

$claude mcp add --transport http google-ads "https://mcp.apify.com/?tools=atkvn/google-ads-transparency-scraper" --header "Authorization: Bearer YOUR_APIFY_TOKEN"

The tool is called atkvn--google-ads-transparency-scraper and takes the same input as described above. It starts a normal run, so pricing and limits are the same. Tested on 21 Nov 2026 with two calls through the MCP server (shopify.com, 3 ads; nike.com with platform: youtube, 5 ads): both succeeded in under 6 seconds and were charged per ad as usual.

Pricing

Pay per result, two events:

EventWhen it is charged
adOnce for every ad row written to the dataset.
ad-detailOnce for every ad whose detail record was loaded (includeDetails). Not charged when the lookup fails.

Notice rows, duplicates, retries and — in monitor mode — ads skipped as known or older cost nothing. If you set a maximum charge for the run, the Actor checks the remaining budget before each ad and stops cleanly when it is used up — it never does billable work it cannot deliver.

Rate limits, speed and reliability — read this before a large run

  • Listing ads is fast and light. One request returns up to 100 ads. In testing, 250 Nike ads took 4 requests and under 4 seconds.
  • No paging ceiling found. On 21 Nov 2026 on the Apify platform, Nike in the US (Google reports 7,000–8,000 ads) was listed twice: 2,500 ads in 26 pages and 90 seconds, then with no limit 2,997 distinct ads in 30 pages and under 3 minutes, containing all 2,500 from the first run. The second run stopped only because it reached the run's maximum charge, not because Google stopped returning pages. For a full export of a large advertiser, set the maximum charge high enough (about 1.50 USD per 1,000 ads without details).
  • Details are slow on purpose. Each ad needs its own request. In testing from a single IP address without a proxy, Google answered about 60 consecutive detail lookups at one per second and then started redirecting every request to its "unusual traffic" page; the address was still refused 42 minutes later, when testing stopped. The Actor therefore defaults to one detail request every 4 seconds, which means roughly 7 minutes per 100 ads with details. Whether that pace avoids the limit indefinitely has not been established.
  • What happens when Google rate limits a run: without a proxy the Actor waits and retries twice, then stops, saves what it has, writes one blocked notice row and marks the run as failed with a clear message. It does not solve CAPTCHAs. Resurrecting the run later continues from the saved position without re-charging or duplicating ads. With plain Apify Proxy (the default) the Actor connects directly first and moves to the proxy only after Google refuses the direct connection, then retries each further refusal through a fresh proxy address. On 21 November 2026 Google refused Apify datacenter proxy addresses on 4 of 6 test runs while direct connections from the same platform answered, so the proxy is a fallback, not the first route. Choosing a proxy group or your own proxy URLs makes every request go through the proxy.
  • Transient errors (timeouts, HTTP 5xx) are retried up to 5 times with exponential backoff.
  • No duplicates. Each query keeps a set of the ads it already wrote; an ad is never written twice for the same query, including after a resume. The same ad can appear under two different queries (for example a domain and that advertiser's ID).

Honest limits

  • Ad copy is usually not available as text. Google serves most text ads in the Transparency Center as a picture of the ad. In a test run of 61 Shopify ads with details, 44 were text ads and only 1 had copy that could be read as text; the rest have an imageUrl you can view or pass to OCR / a vision model. adText is null in those cases — it is never guessed.
  • landingDomain is filled for domain queries (100% in testing) and is usually null for advertiser-ID and name queries, because Google's list does not include it there.
  • Impressions are ranges, published by Google only for ads shown in the EU and a few other regulated regions. No spend, clicks, targeting keywords or conversion data exist in the Transparency Center.
  • Image ads that have no static rendering only expose a preview script URL in previewUrl; open transparencyUrl to see them.
  • Platform filter (Search / YouTube / Shopping / Maps / Play) only covers ads shown since 4 Sep 2023, because that is all Google applies it to. Example from 21 Nov 2026: nike.com in the US gave 8,000–9,000 matching ads on all platforms and 3,000–4,000 with platform: youtube.
  • advertiserVerified is read from a flag in Google's response that matched the site's "verified" badge on the advertisers checked; Google does not document it.
  • Totals are approximate. Google reports the number of matching ads as a range (for example "7,000–8,000"); the log shows it at the start of each query.
  • This is an unofficial tool. It depends on the Transparency Center's internal endpoints, which Google can change without notice. If a field starts coming back empty, open an issue.

FAQ

Do I need a Google account or API key? No. The Transparency Center is public.

Can I search by keyword, like "running shoes"? No. The Transparency Center is organised by advertiser and by landing domain, not by ad keyword. Query the brands or domains you care about.

Why does a domain query return advertisers I do not recognise? A domain query returns every advertiser whose ads point at that domain: regional legal entities, agencies, resellers, affiliates. Filter by advertiserId or advertiserName afterwards, or query the advertiser ID directly.

How far back does the data go? As far as the Transparency Center shows for that advertiser. Use dateFrom / dateTo to narrow it.

How do I monitor a competitor over time? Turn on monitor mode (onlyNewAds with a monitorName, see above) and schedule the Actor, for example weekly: each run returns only ads first shown since the previous run, and ads it skips are not charged.

Can I get more than maxAdsPerQuery allows? Set it to 0 for no limit. For advertisers with tens of thousands of ads, split the work by region, format, platform or date range, and set a maximum charge.

Can I run it on my own machine? Yes:

python -m venv .venv && .venv/bin/pip install -r requirements.txt
echo '{"queries": ["nike.com"], "region": "US", "maxAdsPerQuery": 50}' > input.json
.venv/bin/python -m src --local --input input.json --output ads.jsonl

--local stores the dataset and resume state under ./storage and needs no Apify account. Add --resume to continue an interrupted local run.

Is this legal? The Actor only reads data Google publishes for public transparency, without logging in. You are responsible for how you use the output; advertiser names can be names of individuals, so handle the data in line with the privacy rules that apply to you.

Development

.venv/bin/pip install -r requirements-dev.txt
.venv/bin/python -m pytest # offline tests, real captured responses as fixtures
RUN_LIVE_TESTS=1 .venv/bin/python -m pytest tests/test_live.py # one small live test (about 8 requests)