Google Ads Library & Microsoft Ad Library

Competitors' Google ads (Ads Transparency Center) and Bing ads (Microsoft Ad Library) as JSON for AI agents: ads by domain, name or ID with shown dates and ad wording, via Apify's MCP server.

Hosted MCP Server

npx add-mcp 'https://mcp.apify.com/?tools=fetchwright/google-ads-transparency-scraper,fetchwright/microsoft-ads-library-scraper'

Installs into Claude Code, Codex, Cursor and more

Documentation

Google Ads Transparency Scraper API: Competitor Ads Library (fetchwright/google-ads-transparency-scraper) Actor

Competitor ad tracker for Google Ads Transparency via API: ads by domain, name or ID with first/last shown dates, impression ranges (mainly EU), optional own-ads filter and new-ad alerts to Slack. Summary says which competitors were read in full. $0.60/1,000 ads on Free; ad text +$0.15/1,000.

Pricing

from $0.45 / 1,000 ads

This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events. Since this Actor supports Apify Store discounts, the price gets lower the higher subscription plan you have.

Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event

What's an Apify Actor?

An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server.

Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md.

How to integrate an Actor?

If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready.

Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run.

For examples already wired to this Actor's own input schema, see the API section below.

Each client library has reference documentation the quickstart does not restate: JavaScript/TypeScript (npm install apify-client) and Python (pip install apify-client).

README

What does Google Ads Transparency Scraper do?

Google Ads Transparency Scraper turns your competitors' Google ads, as Google publishes them for the advertiser accounts you choose, into one clean table for a base price of $0.60 per 1,000 ads on Apify's Free plan ($0.45 to $0.55 on paid plans), and can watch those accounts for new ads. Optional ad text adds $0.15 per 1,000 saved ads with text on any plan. Each run states, per advertiser, whether its list was read to the end, stopped at your maximum, or cut short. Enter a domain, advertiser name or advertiser ID and get one row per ad from the Google Ads Transparency Center, across Google Search, YouTube, Display, Shopping, Maps and Play: its format, first and last shown dates, preview and link, and on request its wording, video, countries and impressions. Try the prefilled example to request up to 100 of Nike's ads.

Use it as a competitor ad tracker and a Google Ads Library API: schedule it to monitor your competitors' Google ads and get new-ad alerts in Slack or a webhook, or call it as a Google Ads Transparency API from your own code.

Your first check

  1. Open the example task Monitor your competitors' new Google ads, or paste your own competitors' domains (such as nike.com), names or advertiser IDs.
  2. Turn on Only ads not returned in earlier runs (the example task already has it on), run once to get and remember their current ads, then schedule it daily: each later run returns only ads it has not returned before.
  3. That task reads up to 50 ads for each of 3 domains, 150 ads, with ad text: about $0.11 on the Free plan and $0.090 on Business (ads at $0.60 / $0.45 per 1,000 plus ad text at $0.15 per 1,000 ads with text), plus Apify's start fee of $0.00005. Without ad text the same 150 ads cost about $0.09.

One ad, as saved by a run on 15 November 2026 (abridged):

{
  "advertiserName": "Asana, Inc.",
  "advertiserVerified": true,
  "format": "TEXT",
  "firstShown": "2026-11-10T22:08:47.000Z",
  "lastShown": "2026-11-15T09:57:14.000Z",
  "daysShown": 6,
  "domain": "asana.com",
  "url": "https://adstransparency.google.com/advertiser/AR06554691289986629633/creative/CR11244337948453240833"
}

Built for agencies, growth marketers, sales teams and analysts, developers and AI agents (how each uses it). The Transparency Center website is made for looking up one advertiser at a time; this Actor turns it into data you can schedule, compare and export.

Google Ads Transparency Scraper: competitors' Google ads in one table

What you getDetails
๐Ÿ”Ž Search by domain, name or IDnike.com, Nike, AR16735076323512287233, or a Transparency Center URL
๐ŸŒ Filter by countryAny two-letter country code, or all countries
๐ŸŽฌ Filter by format and platformText, image or video; Google Search, YouTube, Shopping, Maps or Play
๐Ÿ“… Filter by dateKeep ads whose running period overlaps your dates
๐Ÿ†• Get only new adsScheduled runs return just the ads not returned before
๐Ÿ”” Get an alert in Slack or DiscordPaste a webhook address and each run posts what it found: ads per advertiser, the first three with headline and link. The run summary says whether the alert was delivered
โœ… Know what you gotFor each advertiser the run summary says whether its list was read in full, stopped at your maximum, or cut short
๐Ÿงฉ Get variations, countries, impressions and platformsEvery variation of an ad, where it ran with first and last shown dates, the impression range Google publishes and the Google platforms it ran on
โœ๏ธ Get the ad textHeadline, description and display URL of Search text ads: Google's exact text where its preview carries it, otherwise read from the ad image; description, call to action and destination URL of video and app ads; product, price and merchant of Shopping ads
๐Ÿท๏ธ See offers, and what is new since the last runWith ad text on, each ad lists the offers its wording states (free trial, discount, promo code, price and more), each with the exact phrase; a scheduled watch reports the offers newly observed since its previous run and adds them to the alert
๐ŸŽฌ Get the videoThe YouTube video ID, link and thumbnail of video ads, the video's length where Google states it, and the app ID of app ads
๐Ÿชช Get the advertiser's country and verificationWhere the advertiser is registered, and whether it is verified, not verified or unknown
๐ŸชŸ Add Microsoft Advertising ads (EEA)Set Ad network to Microsoft Advertising for the Bing ads Microsoft publishes for the European Economic Area, with wording, landing page, images and, on request, more. Same price per ad
๐Ÿข Check who advertisesAdvertiser summary mode: one row per company with whether it runs Google ads, roughly how many, which accounts and which formats
๐Ÿ’พ Export anywhereJSON, CSV, Excel, or through the API

No Google account, cookies or browser are needed. The Actor reads only what Google publishes. It is an independent tool, not an official Google API, and not affiliated with, endorsed by or sponsored by Google.

Why choose Google Ads Transparency Scraper?

  • Your limits hold. The maximum ads per advertiser is a hard cap, and a maximum charge you set in Apify stops the run before it is exceeded (details). Every run's summary says how fully each list was read, so a short result is not mistaken for a complete one.
  • Counts you can check against Google's. In a test on 10 November 2026 on four advertisers in the US, the number of ads returned equalled Google's exact count for one advertiser (69 of 69) and fell inside the range Google publishes for the other three: 241 (Google: 200 to 300), 1,249 (1,000 to 2,000) and 123 (100 to 200). A count inside a range is consistent with a full list but does not prove it; the run summary's listReadToEnd says which lists were read to the end. The 1,682 ads took 64 seconds without ad text or details.

How much does it cost to scrape Google Ads Transparency Center?

$0.60 per 1,000 ads on the Free plan, $0.45 to $0.55 on paid plans. You pay per ad saved, the run's platform usage is included, and there is no monthly rental.

Your Apify planPrice per 1,000 ads
Free$0.60
Starter (Bronze)$0.55
Scale (Silver)$0.50
Business (Gold) and above$0.45

A run also costs a start event of $0.00005 per GB of run memory (rounded up, at least one GB). Impressions, variations and the advertiser summary are included in the base price. Ad text +$0.15 per 1,000 ads with text, on every plan. The add-on is charged once per saved ad with non-empty wording successfully read, with Include ad text on. Ads without read wording and runs with that switch off have no ad-text charge.

Never charged: with one run of a new-ads watch at a time, ads it has recorded as returned (two overlapping runs of the same watch, or a run killed before it saved its memory, can return and charge an ad again), ads filtered out by your dates, and what was not read of targets that failed or did not finish (listed under failedTargets and unfinishedTargets in the run summary, never shown as "no ads"; the ads saved for them before they stopped are charged).

Companies without ads: by default they get no row and no charge, and are listed under noAdsTargets in the run summary. In advertiser summary mode with also save companies without ads on, each gets one row, charged as one row.

An ad whose extra details could not be read is still a saved ad: it is charged and carries detailsError.

What typical jobs cost:

JobAdsFreeStarterScaleBusiness and above
One competitor's current ads in one country100$0.060$0.055$0.050$0.045
Ten competitors, 200 ads each2,000$1.20$1.10$1.00$0.90
Daily monitoring of five competitors, about 10 new ads a day (after the first run, which returns the current ads up to your cap)300 a month$0.18 a month$0.165 a month$0.15 a month$0.135 a month
Checking which of 1,000 companies advertise on Google1,000 rows$0.60$0.55$0.50$0.45

These base costs exclude the optional ad-text add-on (+$0.015 for 100 ads with text) and the start event above.

Apify's free plan includes $5 of credit each month, which is about 8,300 ads with this Actor without the ad-text add-on.

Caps and a maximum charge: see Can a run cost more than I set?

How does the price compare?

Prices per 1,000 ads of the five most-used Google Ads Transparency actors on the Apify Store plus every other with more than 10 monthly users, plan by plan, from each actor's pricing on the Apify API on 14 November 2026, listed by monthly users.

ActorMonthly usersFree (base)Starter (base)Scale (base)Business and above (base)Extras per 1,000
This Actor$0.60$0.55$0.50$0.45Ad text +$0.15 per 1,000 ads with text, on every plan; details and advertiser summary included
Most-used1,652$2.00$2.00$2.00$2.00Details +$3.00, ad text +$2.00, advertiser rows $3.00, on every plan
Second883$1.90$1.60$1.30$0.90Video download +$4.00 (Free, Starter), +$2.00 (Scale and above); start event $0.00005 per GB of run memory
Third622$1.50$1.00$0.90$0.80None; a start event of $0.001 per GB of run memory
Fourth166$2.00$2.00$2.00$2.00None
Fifth159$1.48$1.24$1.00$0.77Start event $0.00005 per GB of run memory; its listing says platform usage comes on top
jaybird71$0.45$0.42$0.38$0.35Landing page +0.05; start event $0.00005
automation_craft21$0.30$0.30$0.27$0.24Start event $0.0025
maximedupre20$0.90$0.90$0.80$0.65None
jmlp15$0.17$0.17$0.15$0.15None
t2solutions12$0.50$1.10$1.00$0.90Start event $0.00005

The plan columns compare base per-ad prices; this Actor's Extras column shows the optional ad-text add-on.

On every plan this Actor's base price is below the five most-used actors' and maximedupre's; jaybird, automation_craft and jmlp charge less per ad on every plan, and t2solutions on the Free plan (competitors checked 14 November 2026; this Actor's price from 17 November 2026).

Prices change; check each listing. We compared prices only, not results.

How do the features compare?

What the listings of the first four actors in the price table say, read on 11 November 2026. We read their listings and did not run them. "Not described" means the listing does not mention it, not that the actor cannot do it.

This ActorMost-usedSecondThirdFourth
Only the new ads on scheduled runsYesYesNot describedNot describedNot described
Wording of Search text adsOptional, +$0.15 per 1,000 ads with text: exact where Google publishes it, otherwise read from the imageOptional, charged extraOptional, read from the imageNot describedYes, with each variation
One row per company: does it advertise, and roughly how muchYesYesNot describedNot describedNot described
Slack or Discord alert posted by the run itself, with no integration to set upYesNot describedNot describedNot describedNot described
Says whether each advertiser's list was read to the endYes, for every advertiserSays which target stopped part-way, and after how many adsNot describedNot describedNot described
Report you can forward to a clientYes, freeNot describedNot describedNot describedNot described

Apify's own Slack, webhook and email integrations work with every actor on the Store, these four included.

Where the others do more: the second and fourth can save video and image files for you, the second can limit a search to political ads, the third states up to 50,000 ads in one run, and the fourth accepts Google account cookies to reach age-restricted ads. This Actor saves no media files, reads political ads only for the 12 countries of Google's political report, has been measured up to 1,249 ads for one advertiser, and never uses a Google account. Listings change; check each one before you decide.

How do I scrape Google Ads Transparency Center?

How to scrape Google Ads Transparency Center

Create a free Apify account, open the Actor, add domains or advertiser names, pick a country (and a format if you like), click Start, and download the results as CSV, Excel or JSON or fetch them through the API. For monitoring, see Your first check; Apify's video shows how to schedule a run:

https://www.youtube.com/watch?v=1jI7WcVQmwM

โฌ‡๏ธ Input for Google Ads Transparency Scraper

FieldWhat it does
Advertisers to look upDomains, names, advertiser IDs or Transparency Center URLs. A domain also returns ads of other advertisers that Google links to it; use an advertiser ID for one exact account.
Country where the ads were shownCountry name in English or the country's own language, or a two-letter code, such as US, GB, United Kingdom, Germany or Deutschland, or ANYWHERE. A name read this way is noted in inputNotes; an unreadable value stops the run before any ad is read or charged, and the status message names it and the accepted forms.
Countries (several in one run)Optional list (US, GB, DE) that replaces the single country; Google ads only. Each ad is saved and charged once, and Maximum ads per advertiser applies across the countries together, so the run costs no more than with one country (details).
Ad formatAll, text, image or video.
Maximum ads per advertiserCaps results and cost. Default 100.
Maximum ads per runCaps the whole run, across all advertisers, so a long list cannot multiply the cost. Empty by default: no run-wide cap.
Continue from the last runOff by default. On: a run that stops before every target is finished keeps its place for the next run with the same input (details).
Shown on or after / on or beforeKeeps ads whose running period overlaps the dates. Dates such as 2026-10-01, 01.10.2026 or 1 Oct 2026, or a time before the run such as 30 days, 2 weeks or 3 months. A value that cannot be read (for example 11/09/2026, where day and month are unclear) stops the run before any ad is read or charged, naming the field, the value and the accepted forms.
Political ads onlyReads Google's political (election) ads instead of all ads, for one country Google's political ads report covers. Countries, fields and price: Political ads.
Only ads not returned in earlier runsNew-ads-only mode for scheduled monitoring.
On a watch's first runReturn the current ads (up to the cap) or only remember them, so the next run starts from new ads.
What to returnAds (one row per ad) or Advertiser summary (one row per company).
Include variations, countries, impressions and platformsAdds each ad's variations, countries with dates, impression range and platforms. Slower: one extra request per ad (How fast is it?).
Include ad textOptional: +$0.15 per 1,000 ads with text on any plan. Adds the ad's wording, and for video ads the YouTube video. One extra request per ad (How fast is it?).
Ad text for the first ads onlyWith ad text on, reads the wording for at most this many ads per advertiser (the ads Google lists first) and saves the rest without it. 25 gives a wording sample of each advertiser in about a minute, with the full list still saved. Empty reads every ad.
Monitor nameOptional name that keeps several new-ads-only watches apart.
Google platformsKeep only ads that ran on Search, YouTube, Shopping, Maps or Play. Works for every country at no extra cost.
Read targets from a datasetTake domains, names or advertiser IDs from another run's dataset.
Include advertiser country and verificationAdds advertiserCountry, advertiserVerified (verified, not verified or unknown) and advertiserLegalName to ad rows and advertiser summary rows. On by default, no extra charge.
Advertiser identity lookups per runHow many advertiser accounts get that lookup. Default 25; raise it above the number of advertiser accounts in your run to cover every account you named (accounts found by a name search already carry it). Under a domain, only the domain's own advertiser accounts are looked up (or, with none, its most frequent account).
Domains: only the domain's own advertiser accountsOff by default. On: a domain target saves and charges only ads from the domain's own advertiser accounts (isDomainAdvertiser true); the others are left out and not charged (details).
Send an alert to Slack, Discord or a webhookPosts a short message when the run saves ads (details). No extra charge.
Ad networkGoogle (default) or Microsoft Advertising, for Bing ads shown in the European Economic Area.
Advertisers per name searchA name can match several advertiser accounts; the one with the most ads is used. Default 1.

Google connection settings

On Apify, omitting proxyConfiguration uses an Apify datacenter session shared by Google data reads. Set "proxyConfiguration": {"useApifyProxy": false} to go direct. Explicit custom or Apify proxy settings are respected; omitted local settings stay direct. Ad previews, images and ad-text fetches keep their existing route. A refused first connection is retried once (see If Google asks for verification under Limits). There is no automatic residential fallback. A new session ID does not guarantee a different exit IP.

Pasted Transparency Center links

A link copied from the Transparency Center keeps its own filters: its country (region=DE), format (format=VIDEO) and dates (start-date, end-date) apply to that link's advertiser or domain only, so two links with two countries are each read for their own country, and searchRegion on each row says which.

  • A link's filter is used only where you left the field unset (country ANYWHERE, format all, no date). A country, format or date you set applies to every target, and the run notes where a link asked for something else.
  • A link's filter only narrows what is read; it never adds countries, formats or dates.
  • A link to one ad returns that ad, with its country used for the lookup and in searchRegion; in advertiser summary mode, too, only a link's country is used. In both, its format and dates are not applied, and the run says so.
  • The same advertiser pasted with different filters (for example two countries) is read once per set of filters, each up to Maximum ads per advertiser. An ad found for two targets is saved and charged once, under the first of them. Two targets that lead to the same advertiser account with the same filters (its ID and a name, or two names) read it once, in either result type: the second is listed under sameAdvertiserTargets in the run summary, not as a failure, and nothing more is charged for it.
  • A link with topic= other than all stops the run before any ad is read (this Actor cannot filter by topic); nothing is charged apart from the start event.
  • platform, preset-date and unknown parameters are ignored with a note; set Google platforms or the dates instead.
  • A new-ads watch keeps what it has already seen when a link's filters or a monitor name read from another field give it a new memory key, with the same limits as under Never charged in the pricing section.
  • Every decision is in the log and in inputNotes of the run summary.

Coming from another Google Ads scraper?

An input saved for some of the most-used Google Ads Transparency actors (fields read on 11 November 2026) can be pasted here: the fields below are read, a few are not applied, and two filters this Actor cannot enforce stop the run before any ad is charged. A field you set under this Actor's own name, to a non-default value, wins. The log and inputNotes say how each field was read and what to set instead; a filter that stops the run is named in the status message.

Field you haveWhat happens here
domains, advertiserIds, queries, searchTerms, searchQuery, startUrls, urlsRead as targets. Links keep their own filters (above)
maxAdsPerSearch, maxAds, resultsLimit, limitPerSource (a positive number)Read as maxAdsPerTarget
maxResults, maxItems (a positive number)Read as maxAdsPerTarget, and also as maxAdsPerRun, because those actors count them over the whole run
startDate, shownAfterRead as dateFrom
endDate, shownBeforeRead as dateTo
adFormat, adFormats: ["Video"] (one format)Read as format
platform, and platforms spelt SEARCH, google_search or Google SearchRead as platforms
skipDetails: false, includeRegionStats: trueRead as includeDetails: true
monitorStoreName (a name other than ads-transparency-monitor)Read as monitorName
maxAdvertisersPerQuery, maxAdvertisersPerTerm (above 1)Not applied, with a note: one account is read per name; set maxAdvertisersPerName (up to 10) for more
resultsLimit: null, maxResults: 0 ("no limit")Not applied, with a note: maxAdsPerTarget (default 100) applies
downloadMedia: true, shouldDownloadAssets: true, shouldDownloadPreviews: trueNot applied, with a note: no files are saved; rows carry the image, preview and video links
cookiesNot applied, with a note: this Actor reads only what Google shows without an account
urlsFromFile, requestsFromUrl (a link list inside startUrls)Not applied, with a note: paste the links themselves into targets
monitorStoreName: "ads-transparency-monitor" (with new-ads-only on)Not applied, with a note: it is another actor's default name; set monitorName to keep the watch apart
adFormats: ["Text", "Image"] (two of the three formats)Stops the run before any ad is read; only the start event is charged. Set one format, or run once per format
topic: "POLITICAL", political: trueRead as politicalAds: true (Political ads)
topic: "COMMERCIAL" (any topic other than all or political)Stops the run before any ad is read; only the start event is charged. Remove the field to read all ads

region, resultType, includeDetails, includeAdText, onlyNewAds, dateFrom and dateTo have the same names here.

What differs on five problems raised on those actors' Issues tabs (read 11 November 2026: a limit that did not hold, fewer ads than expected, ad wording, no impressions, a run that ends without data): see Can a run cost more than I set?, The run summary, Ad text and Limits. One real run:

  • What a real run returned. 12 November 2026, nike.com, ad text and details on: 30 ads (25 Search text, 3 image, 2 video): wording for 25 (20 read_from_image, 5 found), 3 not_in_preview, 2 image_unreadable (the run summary's adText block counts ads with a preview, with wording, and why wording was missing). One image-read description (confidence 89) contains "Return policy 4.8 ke & wee & (61)": the kind of small error image-read text can carry.

Defaults that can surprise:

  • A name search reads one advertiser account (which) unless you raise maxAdvertisersPerName (up to 10).
  • Each advertiser stops at 100 ads unless you raise maxAdsPerTarget.
  • Ads from all countries are returned unless you set region or a pasted link carries a country.

โฌ†๏ธ Output from Google Ads Transparency Scraper

Output table: one row per ad

One item per ad:

{
  "advertiserId": "AR16735076323512287233",
  "advertiserName": "Nike, Inc.",
  "advertiserCountry": "US",
  "creativeId": "CR04360763343116435457",
  "format": "IMAGE",
  "firstShown": "2023-11-17T02:43:09.000Z",
  "lastShown": "2026-11-09T19:38:14.000Z",
  "daysShown": 922,
  "domain": "nike.com",
  "imageUrl": "https://tpc.googlesyndication.com/archive/simgad/3624387990879582559",
  "url": "https://adstransparency.google.com/advertiser/AR16735076323512287233/creative/CR04360763343116435457?region=US",
  "searchTarget": "nike.com",
  "searchRegion": "US",
  "scrapedAt": "2026-11-09T21:00:00.000Z"
}

Variations, countries, impressions and platforms

With Include variations, countries, impressions and platforms each item also has variationCount, variations (each with imageUrl and previewUrl), platforms (such as GOOGLE_SEARCH), impressionsMin, impressionsMax and regions: per country (region), its firstShown and lastShown dates, impressionsMin, impressionsMax and platforms with each platform's impression range.

Google publishes impressions as a range (mainly in the EU, see Limits). A missing lower bound means "fewer than the upper bound"; a missing upper bound means "more than the lower bound". Both are null where Google publishes nothing.

Ad text

With Include ad text each item also has the wording:

{
  "headline": "Nikeยฎ Slides & More - Nikeยฎ Just Do It",
  "description": "Kids slides designed for comfort at every step. Shop now at Nikeยฎ. ...",
  "displayUrl": "www.nike.com/shop/nike-slides",
  "callToAction": null,
  "destinationUrl": null,
  "products": [],
  "adTextStatus": "read_from_image",
  "adTextConfidence": 90,
  "adTextQuality": "clean",
  "adTextJunkTokens": 0
}

adTextStatus says where the wording came from: found (in Google's preview), read_from_image, no_wording (the preview gave a landing page, video or app, but no wording), or why it is missing (image_unreadable: the image was read but gave no wording; image_not_fetched: the image did not download; ocr_unavailable: the image-reading engine could not run; not_read_source_refused: Google had refused the run; only in the run summary's counts, since such ads are held back instead of saved; not_in_preview, image_only, preview_unavailable). Ads beyond Ad text for the first ads only are saved without wording and without adTextStatus.

Image-read wording has the reader's confidence (adTextConfidence) and a quality label: adTextQuality is clean when the confidence is 90 or more and at most one token is something ad copy would not contain (a stray symbol, a short non-word), otherwise noisy, and adTextJunkTokens counts such tokens. Wording found as text is native; without wording, null. On two sample runs (meal-kit advertisers and HubSpot in the US, 16 November 2026), 183 of 309 image-read wordings (59%) came back clean: 183 of the 186 read with confidence 90 or more, and none of the 123 below 90. Clean means this check passed, not a word-for-word match with the image; the label never changes the wording.

For Search text ads whose wording is in Google's preview (found), the sitelinks are Google's exact text too, and sitelinks lists their titles.

Offers and what is new since the last run

With Include ad text on, each ad also has offers, the offers its wording states (no separate charge). A monday.com Search ad in the US ("... Start Your Free Trial Now!") returned "offers": [{ "type": "free_trial", "value": null, "text": "Start Your Free Trial" }].

type is one of free_trial (value: its length, such as 14 days), discount (20%, $50), promo_code (the code), free_shipping, demo, webinar, free_plan ("free forever", "free plan", "free to start"), price (from $9/month, with currency and period), guarantee (30-day money-back, free returns, best price, lowest price), deadline (ends sunday, today only) or free_cancellation. A price cut stated without a figure ("reduced prices", "Prix cassรฉs") is a discount with value null; "deals" or "save" alone are not offers. text is the ad's exact phrase. Phrases are recognised in English, and common forms in German, French, Spanish and Dutch; price guarantees, free cancellation and price cuts also in Italian. Only what the ad says is listed: a number alone is not a price, and "free" alone is not an offer. offers is empty when the wording states no offer or was not read (adTextStatus says which).

The run summary's offers object lists per advertiser the offers observed in this run (observed: type, value, how many ads state it, and one ad as proof with its link and phrase). With Only ads not returned in earlier runs, the watch also remembers the offers it has seen, and the summary says what is new:

  • newlyObserved: offers this watch had not observed in earlier runs, each with the ad that states it; up to three are added to the Slack or Discord alert.
  • noLongerObserved: offers observed in earlier runs that no ad shown since the previous run states.

How to read them:

  • A newly observed offer is not proof of a changed offer: it may have run earlier in ads whose wording the watch had not read.
  • On a watch's first run, and an older watch's first run with ad text, nothing is newly observed: what it finds is the starting point, marked baseline: true.
  • An offer is reported as no longer observed only when the run read the advertiser's list to the end and the wording of the ads shown since the previous run was read. Otherwise noLongerObserved is left out and noLongerObservedReason says why (baseline, list_not_read_to_end or wording_not_read_for_all_ads).
  • Offers can run side by side. A 30-day trial newly observed while a 14-day trial is still observed is two offers, not a change of trial length.

Microsoft Advertising rows carry offers too, read from headline and description, and their run summary lists the offers observed.

Advertiser summary

In Advertiser summary mode each item describes one company instead of one ad:

{
  "searchTarget": "nike.com",
  "hasAds": true,
  "adCountMin": 8000,
  "adCountMax": 9000,
  "advertisers": [
    { "advertiserId": "AR16735076323512287233", "advertiserName": "Nike, Inc." }
  ],
  "formats": ["IMAGE", "TEXT", "VIDEO"],
  "mostRecentAdShown": "2026-11-09T19:38:14.000Z",
  "searchRegion": "US"
}

Companies without ads, and companies that could not be checked, are handled as under Companies without ads and Never charged in the pricing section. With Advertiser summary: also save companies without ads on, every company gets a row with hasAds and a targetStatus of has_ads, no_ads_in_region or no_advertiser_found, so the dataset lines up with your list.

Microsoft Advertising ads (EEA)

Set Ad network to Microsoft Advertising to read Microsoft's ad library instead of Google's: Bing ads shown in the European Economic Area (the 27 EU countries, Iceland, Liechtenstein and Norway). Rows are charged at the same price as Google ads, through the same maximum charge, and the run summary has the same outcome, coverage, failedTargets and noAdsTargets, plus failuresByReason. For example: {"adNetwork": "microsoft", "targets": ["Booking.com B.V.", "4295086664"], "region": "ANYWHERE", "maxAdsPerTarget": 10, "includeDetails": true}.

Microsoft details here are for small runs: with up to 25 Microsoft ads in the run (the example reads at most 20) they are included; on larger runs they are skipped (not charged) and the summary says so, and single-ad lookups, which always include details, run for the first 25 only, whether or not details are switched on. For details on more Microsoft ads use the Microsoft Ads Library Scraper, which includes them at no extra charge.

Targets are advertiser names as Microsoft lists them (often the legal name, such as Booking.com B.V.) or Microsoft's numeric advertiser IDs. A name can match several accounts: the account whose name matches exactly is read first, and the others are listed under otherMatches in the run summary; raise Advertisers per name search to read more. Microsoft cannot be searched by website, so a domain such as booking.com is reported as not usable (not charged) instead of returning nothing.

Country: ANYWHERE (all EEA countries) or one EEA country such as DE, FR or NL. With a country outside the EEA (the form suggests US), the run searches ads shown in any EEA country and says so in the log and the run summary's regionNote. A run stopped for another input problem, such as no target, writes no summary.

Each row has adNetwork microsoft, the usual advertiser, link and search fields, wording (headline, description, displayUrl, destinationUrl), imageUrl and images, and firstShown, lastShown, impressionsRange (such as 100K - 250K) with impressionsMin and impressionsMax, impressionShareByCountry (country, countryName, sharePercent), targeting (type, usedForExclusion), paidBy and restrictionReason.

The fields from firstShown to restrictionReason need Include variations, countries, impressions and platforms (one extra lookup per ad, no extra charge; only on runs of up to 25 Microsoft ads, see above); without it, firstShown and lastShown are null and the others are left out. images lists the images Microsoft gives, and imageUrl is the first. paidBy is filled only when Microsoft names a payer other than the account owner, so null means either no other payer was named or the details were not read (option off, or the lookup failed), not a confirmed payer. It is not the same field as Google's fundedBy, which Google states also when the payer is the advertiser itself.

Not available for Microsoft ads, because Microsoft does not publish it: the ad format (format is always null), ads shown outside the EEA, search by website (above) and exact impression counts (only a range). Not offered for Microsoft ads in this Actor: the advertiser summary mode, new-ads-only monitoring, political ads only, and the format and platform filters (choosing one stops the run with a message, not charged). Microsoft states how many ads an advertiser has (up to 1,000); that Microsoft actor gives the count per advertiser and has new-ads-only monitoring. The client report covers Google ads only (with Microsoft the run says so instead). Ad text options are not needed: Microsoft rows always carry their wording.

Political ads

With Political ads only on, the Actor reads the ads Google lists in its political ads report for the country you choose: AR, AU, BR, CL, GB, IL, IN, MX, NZ, TW, US and ZA, the countries the report lists on 12 November 2026 (the EU is listed without a single country to read it by and is not offered). Any other country, or none, stops the run before anything is read. Each political row has the usual ad fields plus:

  • spendCurrency, spendMin and spendMax: the spend on the ad, as the range Google states, in the report's currency (USD for the US). spendMax is null when Google states only "Over" an amount. spendRanges lists every currency Google states; nothing is converted.
  • impressionsMin and impressionsMax: times shown, as Google states them.
  • politicalRegions: the regions Google lists the ad under (["US"]).
  • With details on (one lookup per ad, no extra charge): targetingAge, targetingGender and targetingLocations, each with the values the advertiser included and excluded as Google states them, or null when it states none. Locations are Google's own numeric location IDs (geoId); only an ID that is a whole country also gets its two-letter code in region, and states, districts and other areas keep region: null rather than a guessed name. Like audience selection, this is the criteria the advertiser chose, not who saw the ad. fundedBy is null when Google's details state no funding name: the US political ads we read on 12 November 2026 stated none.

Political ads cost the same per ad as other ads, under the same caps, maximum charge and new-ads-only mode; a political watch keeps its own memory, apart from an ordinary watch on the same advertiser.

Personal data. Google's political ad data names the advertisers and the organisations that fund the ads, and can name natural persons, such as candidates or individuals who pay for ads. These names appear in the rows (advertiserName, fundedBy, searchTarget), in the run summary and in alert messages; ad text and images can contain names too. That Google publishes them does not end data-protection duties such as those of the GDPR. When you store or use this data, you decide why and how, which usually makes you the controller for that use: you are responsible for a lawful basis, for retention, and for deletion when you no longer need it, including the copies in your datasets, exports and alert channels.

Client report (free)

Client report sample

Open the full sample report, written by a run on 10 November 2026 and unedited.

With Write a client-ready report on, the run also saves REPORT.html, one page to forward or print, and REPORT.md, linked from the run summary under report. It leads with the strongest true finding about the run's advertisers and shows real ads as proof. Every figure is counted from this run's rows and every sentence comes from a fixed template; no AI model writes it. A report is written when the run has ads from two or more advertisers and at least one advertiser account has its whole list read or at least 100 ads saved. A domain's ads are often spread over several accounts, so with domains set Maximum ads per advertiser to 200 or more (a higher maximum can cost more; listReadToEnd in the run summary says which lists were read in full). When no report is written (the option is off, or the run does not have enough data), REPORT.html is a short note and the run summary says why and what to change; in code, report.status is ok only when a report was written. It never changes the run's results or charges.

For a report that quotes real ads, also switch on Include ad text with Ad text for the first ads only at about 25 (about a minute or two more per advertiser by the pacing estimate).

Other fields and the run summary

  • With details on, each ad also has fundedBy: who funded the ad, as the ad page states it ("Ad funded by"), which can differ from the advertiser account (an agency or another group company: one of Zalando SE's ads in Germany states DQ&A Technology NL BV). Same detail lookup, no extra request or charge; null without details, when the details could not be read, or when Google states none.
  • EU ads: topic. With details on (same lookup, no extra request or charge), each ad also has topic: the subject EU ad pages show as "Topic (labeled by Google)", in English, for example Business & Industrial, Apparel or Finance (22 topics), with Google's numeric topicId; topicSource is google when Google labeled it, and topicSourceCode keeps Google's code as sent. Google publishes it mainly for ads shown in the EU; elsewhere, or when the details could not be read, the fields are null (left out with details off). In a German run on 16 November 2026, all 8 detail answers we kept carried a topic. An unseen topic ID keeps its topicId with topic null; the advertiser-provided label has not been seen yet, so its code would show only in topicSourceCode.
  • With details on, each ad also has audienceSelection: for demographics, locations, contextual signals, customer lists and topics, whether the advertiser included or excluded criteria (included, excluded, included_and_excluded, not_used). Google publishes it for a minority of ads, mainly those shown in the EU. As Google says on the ad page, this describes the criteria the advertiser chose for its most recent campaign, not who saw the ad; a category is not_used when Google lists it without criteria or leaves it out, and the whole field is null when Google publishes no audience selection for the ad.
  • The run summary's googleRequests counts every request sent to Google (lookups, ad previews and ad images) and peakRequestsPerMinute the most in any one minute; requests counts lookups only.
  • impressionsMin and impressionsMax are what the ad page calls "Times shown". When Google has not published them yet, timesShownAvailableFrom gives the day it says they will be (for example 2026-11-27), and both are null; nothing is estimated.
  • googleCustomerId, googleCreativeId, adGroupId and versionId come from Google's preview link when it carries them; imageWidth and imageHeight for image ads.
  • advertiserCountry is the advertiser's registered country. advertiserVerified is true when the Transparency Center shows the advertiser as verified, false when it shows "hasn't verified their identity yet", and null when unknown; nothing is guessed. Both come from Google's advertiser search, one lookup per advertiser, within Advertiser identity lookups per run (input table). Rows of other advertisers linked to a domain keep null unless the run already found the account (the summary counts them under identity.otherAdvertisersNotLookedUp; give such an advertiser's ID as a target to get its identity). They are unknown for advertisers beyond that number and for advertisers the search does not return.
  • isDomainAdvertiser and linksToDomain (Google ad rows of a domain target only): see Ads from other advertisers under a domain.
  • advertiserLegalName is the name a verified advertiser verified with Google, under which the Transparency Center lists its account (Nike, Inc.). It is null when the advertiser has not verified, when its status is unknown (beyond the lookup limit, or not returned by Google's advertiser search), and on Microsoft rows. It comes from the same lookup as the country, at no extra request or charge. In advertiser summary mode each account in advertisers carries it too, and a summary row for an advertiser ID or name has it at the top level.
  • With ad text on, video ads carry youtubeVideoId, videoUrl, videoThumbnailUrl and sometimes videoDurationSeconds; app ads carry appId and appStoreName; ads for a business location carry businessName, businessCategory and advertiserDomain.
  • daysShown is the number of days Google reports the ad as shown.
  • The earliest firstShown we have seen from Google is 25 October 2021, and long-running ads of different advertisers share exactly that date. Read a firstShown of 2021-10-25 as "on or before", and daysShown for those ads as a minimum.
  • imageUrl is set for image ads and for text ads that Google renders as an image. Video ads and some responsive ads have a previewUrl instead.

The run summary

Each run stores a summary under the key SUMMARY in its key-value store: ads per advertiser, targets that failed (failedTargets, charged only for ads saved before the failure) and targets it did not finish (unfinishedTargets). targetsLeft lists every target not read in full, with stoppedBy (for example source_refused when Google refused or rate-limited the run, time, budget or max_ads_per_run) and rowsSaved; a target not started has rowsSaved 0 and was not charged. A run that reaches your maximum charge or its timeout stops cleanly and keeps every saved row. Two fields say how the run went:

FieldWhat it says
outcome: complete, partial or failedWhether the run did what you asked. complete means every target finished, even if it stopped at your Maximum ads per advertiser. The status message of a partial run starts with "Partial:".
maxPossibleChargeUsdThe most the run could cost on its price, worked out before anything was read: the start event plus your caps times the base price per ad and, with ad text on, the ad-text add-on. A ceiling, not an estimate; also in the run log. Empty when the price was not available.
coverage: complete, capped, partial or unknownHow much of each advertiser's list was read. complete: read to the end (for a new-ads watch, up to where an earlier full read had reached). capped: stopped at your maximum. partial: stopped by your maximum charge, the timeout or an error. unknown: none of these can be stated.

Each entry under targets has its own coverage, and the run's coverage is the least complete of them. So a run can be outcome complete with coverage capped: it did what you asked, and Google has more ads than your maximum let through (raise it to read the rest). Each entry under targets also has listReadToEnd, stoppedBy when something stopped the list, and Google's approximate total for the search with your region, format, platform and date filters: googleAdCountMin and googleAdCountMax (null when Google gives no bound). Compare it with ads to see how much a cap left unread; on an only-new run it still describes the whole matching list, not the new ads. Advertiser summary mode returns Google's range for the list in the selected region, without format, date or platform filters, as adCountMin and adCountMax.

With ad text on, the summary also has offers. It also has identity (rows with and without the advertiser's country and verification, advertisers looked up, and stoppedBecause when the lookup limit was reached) and notification (see below).

Alerts to Slack, Discord or a webhook

In Slack, the notification line names the advertisers with new ads and how many ("nike.com 3 new (text, newest Nov 15)"); the message lists up to five advertisers, most new ads first, with formats, first-shown dates and up to three ads (with wording as links when Include ad text is on), then a link to the run's dataset. A watch's first run says "Watch started" with the number of ads running instead of calling them new, an ad first shown more than 14 days before the run is marked "older ad, newly seen", and an advertiser that was not read to the end is shown as "not fully read this run" rather than as having nothing new. Discord and other addresses get the same content as plain text. Turn on Alert even when nothing was saved so a quiet week still posts one line.

The alert is sent after every row is saved, best effort: a receiver that is busy or briefly failing is tried up to three times (one that does not answer in time is not tried again, so the same alert is not posted twice), and a failed alert never changes the saved rows or the charge. The run summary's notification object says what happened:

  • status: sent, failed, not_configured (no address given) or nothing_to_send (no ads saved and Alert even when nothing was saved off)
  • httpStatus: the receiver's HTTP answer, or null; error: why it failed, or null; attempts: how many times it was tried

If an alert fails, the ads are still in that run's dataset (open the run or read it through the API) to send on yourself. In new-ads-only mode the ads are remembered as soon as they are saved, so the next run does not return them or alert about them again; read them from the run whose alert failed.

Limits of Google Ads Transparency Scraper

  • What Google lists. You get the ads Google publishes in the Transparency Center for the advertiser accounts you choose, within your filters and your maximum; an advertiser there can be verified, not verified, or of unknown status. A company with no advertiser account of its own returns nothing (why).
  • Counts are approximate. Google reports an advertiser's total as a range (such as 8,000 to 9,000 ads), which the advertiser summary returns.
  • Impressions are EU-only. Google publishes impression ranges mainly for ads shown in the EU, and only per country. Ads shown only elsewhere have no impression data there, so those fields are empty.
  • Ad text: exact or read from the image. Some Search text ads come with their wording in Google's preview; for those you get Google's exact text (found). The rest are read from the ad image (read_from_image), which can contain small errors. In a test on monday.com in the US on 11 November 2026, wording was returned for 39 of 40 ads. Each row read from an image has the reader's own confidence score, which is not a measured accuracy.
  • Video ads. With ad text on, in tests on 10 November 2026 the YouTube video was returned for 90 of 117 video ads in the US and 102 of 134 in Germany, across four and five advertisers. The share differs between advertisers. videoDurationSeconds is filled for a minority. The preview link is always returned.
  • Image ads. In the same US test, 53 of 127 image ads had preview data: 50 returned wording (product title and merchant, or a headline) and 3 a landing page only. The other 74 exist only as a picture (you get the image link).
  • Platforms per ad. The platform filter works for every country; the per-country platform breakdown inside ad details is published mainly for ads shown in the EU.
  • If Google asks for verification. If the first Google request on an Apify datacenter session meets a verification page or HTTP 429, the run retires that session, waits 60โ€“120 seconds and retries the same request once on a new session, when the run timeout permits; a longer requested wait ends the run instead, and HTTP 403 is final. Custom proxy settings and direct runs do not get this restart. The restart adds no charge. If that retry fails, or Google later answers with a verification, CAPTCHA or block page under any HTTP status, a web page where data, an ad preview or an ad image was expected, a refusal (HTTP 403), or an HTTP 429 that outlasts its pauses (below), the run sends nothing more to Google for any advertiser, ad details, previews and ad images included. Ads already read whose details, or wording within Ad text for the first ads only, were asked for but not yet fetched are held back (not saved or charged; adsHeldBack under targetsLeft). With Continue from the last run on, a later run with the same input saves them complete without saving or charging the others again; with it off, no place was kept, so a rerun reads those targets again and charges again for the ads it saves (a new-ads watch returns them on its next run, once). Ads past the ad-text limit are saved as usual. A watch whose first run Google refuses keeps that first run open: the next run rereads the lists from the top, skipping without charge the ads already returned, so a watch refused at the same point every run can return nothing until one gets further; space watch runs hours apart rather than minutes. The SUMMARY record gives route (direct, datacenter, or custom / user_proxy for your other proxy settings), sessionsUsed, firstRequestRetried, and reason plus sourceRefused after a final refusal. Safe failure diagnostics (status, content type, length and matched phrases, not the page) are kept in REFUSAL_PAGE and GOOGLE_SESSION_FAILURES. A later ordinary HTTP 429 without a verification page is paused out on the existing session, never on a new one: at most three pauses (2, 4 and 8 seconds) per request, counting every 429 among its attempts; if Google still answers 429 after that, the run stops asking Google, as for a refusal: sourceRefused is "HTTP 429 after 3 pauses" (with "(ad preview)" or "(ad image)" when one of those met it) and the status says Google rate-limited the run; ad previews and images also stop after four 429s in a row across those in flight, before any one has used its three pauses (sourceRefused "HTTP 429 on 4 ad previews in a row", or images; the status says Google rate-limited ad previews or images). A run that Google refuses or rate-limits ends cleanly (status succeeded, outcome partial), also a one-target run refused before its first ad: finished targets stay finished (one that saved its maximum is not reported as failed), and every target not read in full is listed under targetsLeft (the run summary) and named in the status message ("Saved 200 ads from 1 of 11 targets ..."); nothing is charged for what was not read. With Continue from the last run on, the next run reads exactly those.
  • How much one run covers. Google can stop answering a run before all the advertisers you listed are read. In a test on 17 November 2026 (Apify datacenter connection, US, 20 software companies, up to 100 ads each, ad text on for 25 ads per company), Google refused the run after 5 companies: 500 ads saved, 129 Google requests, under 5 minutes. The run then stopped asking Google, and the other 15 companies were listed under targetsLeft in the run summary, not charged. Since this test, a run reads the ad list of each company it reaches before any ad's wording or details: when Google stops a run, it saves from the lists it had read the ads that need nothing more from Google (up to your maximums), holds back the ads whose wording or details were asked for and not read (see If Google asks for verification), and ends partial with those companies under targetsLeft (listsReadFirst in the run summary). This order is not yet tested against Google. For long lists, turn on Continue from the last run and schedule the actor, so later runs pick up where the last one stopped (how). That is one test: a run can cover more or fewer companies.
  • If saving a row cannot be confirmed. Each row is sent once, never resent by itself. If saving a row reports an error, the run checks its dataset: a row that is there counts once (and against your maximum charge); a row that is not there, or a dataset that cannot be checked, stops the run at that row, so no row is saved or charged twice. The run summary describes it under uncertainWrite, and the advertisers not finished are listed under unfinishedTargets.
  • If the platform restarts a run. Apify can restart a run (for example to move it to another server). A restarted run asks for and saves nothing more, so no ad is saved or charged twice; its status message and run summary (restartedByPlatform) say how many rows it saved before. Start it again for the rest: a new run starts from the beginning.
  • Microsoft Advertising ads. EEA only, recorded since June 2023 and sometimes 1 to 3 days behind. Microsoft serves at most about 1,000 ads per advertiser and query; an advertiser at that limit has coverage unknown, not complete, and sourceLimitReached in the summary. On 11 November 2026, Microsoft did not answer a single-country query for a very large advertiser within about two and a half minutes; the Actor now waits at most 60 seconds for a single-country list, then reports the target as failed (source_timeout, not charged) and says to try ANYWHERE. When Microsoft limits how often it is asked (HTTP 429), the run pauses, slows down and continues. When Microsoft refuses a request, answers with a page saying the request is blocked (under any status, 429 included), or keeps limiting the run after two pauses, the run asks Microsoft nothing more and lists the advertisers not finished as unfinished; what was not read is not charged.
  • Offers. Read from the wording only, never from images or videos, and only for phrases the Actor recognises (languages: Offers); an offer worded another way is not listed. Wording read from an ad image can contain small errors, and so can an offer read from it.
  • Speed. Google is read one request at a time (at most 60 a minute), in the same session up to your item, charge and timeout limits; the pause between targets (and between a target's countries) adds about a minute to a list of 20. A target that reaches your maximum asks Google for no further page. Per-minute figures are estimates, not a runtime guarantee: see How fast is it?

Who uses Google Ads Transparency Scraper, and for what?

๐Ÿ†• Growth marketers: daily competitor ad monitoring

Set up the watch as in Your first check, then paste a Slack or Discord webhook address under Send an alert to get a message whenever a competitor launches new ads, or send the rows to email or Google Sheets with an Apify integration.

The first run returns and charges the current ads up to your cap; what later runs charge, and when an ad can be charged again, is under Never charged in the pricing section.

๐Ÿข Sales teams and analysts: qualify a lead list by Google ad activity

Paste your company domains and set What to return to Advertiser summary: each domain that advertises comes back as one row with approximate ad count, formats and most recent ad (domains with no ads; cost: the last row of the job table in the pricing section).

๐ŸŽจ Agencies: creative research, client reports and pitches

Pull the image or video ads of your category's leading brands in one country and sort by daysShown: an ad run for hundreds of days is one the advertiser chose to keep, worth a closer look, but not proof that it performs (Google publishes no clicks or results). Export a prospect's active ads before a pitch, or a client's competitors for a monthly report.

For a weekly competitor brief, schedule a watch with Include ad text on: see Offers and what is new since the last run.

๐Ÿ”— Developers: chain it from a company list to their ads

Run any Actor that produces company domains (or this one in Advertiser summary mode), then start a second run with that dataset under Read targets from a dataset: it lists the ads of every company in the first run's results, with no export or code.

๐Ÿค– AI-agent builders: use it in Claude, ChatGPT or Cursor through MCP

Add this address as a custom connector (MCP server) in your AI app and sign in to Apify; the Actor then appears as a tool the assistant can call:

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

It runs through the Apify MCP server; ask the agent to "get the ads nike.com is running on Google in the US".

How do I get Google Ads Transparency data with Python, Node.js or the API?

More code: this Actor's API tab; a daily new-ads job: Use it in a pipeline.

You need an Apify API token (Settings > API & Integrations in the Apify Console).

Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("fetchwright/google-ads-transparency-scraper").call(run_input={
    "targets": ["nike.com", "adidas.com"],
    "region": "US",
    "maxAdsPerTarget": 100,
    "includeAdText": True,
})
for ad in client.dataset(run.default_dataset_id).iterate_items():  # apify-client 3; on version 2 use run["defaultDatasetId"]
    print(ad["advertiserName"], ad["format"], ad["firstShown"], ad.get("headline"), ad["url"])

Node.js

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_APIFY_TOKEN' });
const run = await client.actor('fetchwright/google-ads-transparency-scraper').call({ targets: ['nike.com'], region: 'US', maxAdsPerTarget: 100 });
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.length, 'ads');

curl, one call

curl -X POST "https://api.apify.com/v2/acts/fetchwright~google-ads-transparency-scraper/run-sync-get-dataset-items" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" -H "Content-Type: application/json" \
  -d '{"targets":["nike.com"],"region":"US","maxAdsPerTarget":20}'

The one-call form waits for the run and returns the ads; it suits runs under five minutes. For a daily job that returns only new ads, add "onlyNewAds": true and a "monitorName". To check which companies on a list advertise at all, set "resultType": "advertisers".

Use it in a pipeline: new competitor ads to Slack with n8n

Runs are billed to your own Apify account at your plan's prices, wherever you start them.

  1. In n8n, install Apify's node (@apify/n8n-nodes-apify) and add an Apify API credential with your Apify API token.
  2. Add a Schedule Trigger that fires once a day.
  3. Add the Apify node with the operation Run an Actor and get dataset, choose fetchwright/google-ads-transparency-scraper, and set the custom input:
    { "targets": ["competitor-one.com", "competitor-two.com"], "onlyNewAds": true, "monitorName": "competitors", "maxAdsPerTarget": 100 }
    
  4. Add a Slack node that posts one message per item, for example {{ $json.advertiserName }}: new {{ $json.format }} ad, first shown {{ $json.firstShown }}, {{ $json.url }}.

What a watch returns, and when an ad can come again: new-ads-only mode; what reaches Slack then depends on your workflow (a failed or repeated n8n step is outside the Actor's control). Checked on 13 November 2026 against Apify's n8n integration documentation (docs.apify.com/integrations/n8n); the input fields and output columns are this Actor's own. Not run in a live n8n account, so tell us on the Issues tab if a step fails.

Use it as a trigger in Zapier, Make or n8n

A saved task with Only ads not returned in earlier runs on works as a "new Google ads from a competitor" trigger: each finished run carries only ads the watch has not returned before (how it remembers).

  1. In Apify Console, save a task with your competitors and a daily Apify schedule. With On a watch's first run set to Only remember, the first run returns nothing and remembers the current ads without charge. Domains: only the domain's own advertiser accounts keeps the watch on hubspot.com's own ads, not those of other advertisers Google lists under the domain:
    { "targets": ["hubspot.com"], "region": "US", "onlyDomainAdvertisers": true, "onlyNewAds": true, "firstRunBehaviour": "remember_only", "monitorName": "new-competitor-ads", "maxAdsPerTarget": 50 }
    
  2. Start your flow on that task's finished run, not on a run your tool starts (Zapier's synchronous run step stops after 30 seconds):
    • Zapier: Apify's trigger Finished task run, then Apify's search Fetch dataset items with the run's default dataset ID.
    • Make: Apify's trigger Watch Task Runs (an instant webhook), then Get Dataset Items.
    • n8n: the Apify node's trigger On new Apify event, watching that task with event type Succeeded, then Get dataset items.
  3. Send the rows (advertiserName, format, firstShown, url) to Slack, Google Sheets, e-mail or your CRM.

For a Slack or Discord alert alone you need no automation tool: see Alerts to Slack, Discord or a webhook. Runs are billed to your own Apify account at your plan's prices. Trigger and step names checked on 17 November 2026 against Apify's Zapier, Make and n8n documentation (docs.apify.com/platform/integrations/zapier, /make and /n8n); not run in live Zapier, Make or n8n accounts, so tell us on the Issues tab if a step fails.

โ“ FAQ about Google Ads Transparency Scraper

Is there a Google Ads Transparency Center API (or Google Ads Library API)?

Google's programmatic access is a BigQuery public dataset (bigquery-public-data.google_ads_transparency_center), queried with SQL, covering ads shown in the European Economic Area and Turkey, while the Transparency Center website searches all regions, so an advertiser can show far fewer ads in the dataset than on the website (Google developer forum). This Actor reads the Transparency Center itself for the region you choose and works as an API (examples).

How fresh is the data?

Every run reads the Transparency Center live. lastShown is the last day Google recorded the ad as shown (usually today or yesterday for running ads); scrapedAt is when the row was collected.

How fast is it?

Google is read one request at a time, 1 to 1.6 seconds apart (1.3 seconds on average, so never more than 60 a minute), with no per-session request cap, plus a pause of 2 to 5 seconds between targets. Lists come 100 ads per request, so runs without details or ad text are barely slower than before this pacing (before it: 600 ads from two advertisers in 58 seconds on 9 November 2026, about 400 to 600 ads a minute on 10 November). With details on, each ad needs one more request: expect about 37 to 45 ads per minute, so 100 ads in about 2 to 3 minutes. With ad text each ad adds a preview or image request, so about 19 to 23 ads per minute with details and text together; Search text ads whose wording is read from the image take about 3 seconds each. Within the default one-hour timeout, a run with details reads roughly 2,200 to 2,700 ads; a longer timeout reads more. These figures are estimates from the pacing (16 November 2026), not a live measurement or a runtime guarantee.

With Microsoft Advertising, allow about 20 seconds per request before source response time and any slowdown. Runs with details are slower because each ad needs an extra request. This pacing is not a promise of a Microsoft quota or a completed run.

Does it return all of an advertiser's ads?

It returns what Google publishes for the accounts you choose, within your filters and maximum; each run says per advertiser whether the list was read in full (coverage). The 10 November 2026 comparison with Google's counts (toggl.com, getharvest.com, pipedrive.com and duolingo.com in the US) is under Why choose. To get them all, set Maximum ads per advertiser high enough and leave the date and format filters empty; run the advertiser summary first for Google's count.

Ads from other advertisers under a domain

Under a domain, Google lists the domain owner's ads and ads of other advertisers it links to that domain. For hubspot.com in the US on 16 November 2026, 58 of the 200 most recently shown ads came from 42 other advertisers, mostly HubSpot customers. Each ad row of a domain target therefore carries isDomainAdvertiser, decided per advertiser account without any extra request:

  • true: accounts named after the domain, or carrying a real share of the ads that link to it. Named after it: the account's whole name, without legal endings such as Inc, LLC or GmbH, spells the domain's name ("Hubspot, Inc." for hubspot.com, "monday.com Ltd." for monday.com; not "HubSpot Partner Agency" or "Apple Tree Dental" for apple.com). Linking to it: the ad's display or landing address is the domain itself or www. of it, and the account has at least 3 ads in the domain's list, or at least 10% of the ads of the accounts matched to the domain, so a brand's legal entity counts (Optimal Performance LLC for factor75.com) and a one-off affiliate does not.
  • null with linksToDomain true: the account's ads link to the domain, but it has too few of them to count as the domain's own. On 16 November 2026, 5 such accounts under hellofresh.com (one ad each) and 1 under homechef.com. The 3 ads are counted as the list is read, so an account's first two rows stay null when its third makes it true; the 10% share is judged only when the run reads the domain's whole list (not after your maximum or another stop, not with dates or platforms set, not in a continued run or a watch, which read part of it; the summary's linksToDomainShare then says so). In both cases the run summary counts that account's saved rows as own.
  • false: the ad's address is on another site (thetutorverse.com under hubspot.com); the account stays false for its later ads without an address.
  • null: not known, for an account whose name does not spell the domain and whose ad has no readable address (ad text off, image ads, or unreadable). Another subdomain (meetings.hubspot.com) decides nothing, because services lend those to their customers. An account seen with both the domain's own address and another site's is null from then on (it is not turned from false to true, or the other way).
  • A subdomain target (store.example.com) counts only that host and www. of it as its own address, and its first label ("store") as its name. Internationalised domains are compared as written, in either form (bรผcher.de or xn--bcher-kva.de).

On the 16 November hubspot.com run this gives 142 true (all of HubSpot's own two accounts), 47 false and 11 null, with no ad put on the wrong side. Domains: only the domain's own advertiser accounts (input table) saves and charges only the true rows. With ad text off, only the account name decides, so a brand whose account name differs from its domain (a parent company, for example) can be left out entirely: use its advertiser ID instead. The run summary's domainAdvertisers counts these per domain target (ownRows, ownAccounts, otherAdvertiserRows, otherAdvertisers, linksToDomainRows, linksToDomainAccounts, unknownRows and, with the option on, rows left out, leftOutLinksToDomainRows among them). Identity lookups under a domain follow the input table and never go to an account known to be another advertiser's. In advertiser summary mode, each account under a domain row's advertisers carries isDomainAdvertiser from the name rule alone (true or null); with the option on only true accounts are listed, and the row and its charge are unchanged.

With the option on: an unmatched ad's wording is read to see its address, but Ad text for the first ads only counts only the ads kept (wording read for ads then left out counts under adText.readForLeftOut, not as the run's ad text). An ad of an account that links to the domain but has fewer than 3 ads so far is held, not saved or charged: it is saved and charged, before the ad that made the account own, when the account reaches 3 ads (or, at the end of a list read in full, its 10% share), within your maximum and maximum charge; otherwise it is left out and not charged, its wording included. While held it does not use that limit; its wording, already read, is kept and counts toward the limit only if the ad is saved (once the limit is used up, it is saved without wording, like any ad past the limit). A target holds at most 200 such ads (or twice your maximum per advertiser, if higher); past that the oldest is left out, not charged, and counted under linksToDomainHeldOverCap. A watch or a continued run, with the option on or off, keeps the accounts that earlier runs found to be own, so their next new ads are kept. A watch remembers other advertisers' ads and accounts without returning or charging them, so its later runs skip them unread. A domain stops after 200 left-out ads (or twice your maximum per advertiser, if higher) without finding its own (stoppedBy left_out_limit), so a list of other advertisers' ads is not read to the end for nothing.

Why does a company return no ads?

Google lists ads by advertiser account: a company with no account of its own there, or one that advertises under another legal entity or agency, returns nothing for its name. Try the domain instead.

A name returned ads from the wrong company. Why?

A name can match several advertiser accounts (regional entities, agencies, similar names). With a country chosen, the Actor checks the five largest matching accounts and reads the one with the most ads in that country (a brand's US account for the United States, not a bigger account that advertises elsewhere); with all countries, the one with the most ads overall. Each ad row carries its advertiserId and advertiserName, and the run summary lists the other matching accounts (otherMatches, with country, verification and ad count range), so you can check the choice. For one exact account use the domain or advertiser ID; raise Advertisers per name search to read several.

How accurate is the ad text?

Sources (found, read_from_image) and the latest measurement: Ad text: exact or read from the image in Limits. In a test on 10 November 2026 on 52 text ads from four advertisers, text was returned for 47; six of those had visible defects such as stray characters or extra callout text, and ten of the other 41, compared with the ad image word for word, all matched. Those ten were chosen after the visibly defective ones were set aside, so they are not a measure of overall accuracy. Treat image-read text as a close transcription, not an exact copy. For video and app ads the wording comes from Google's own preview data and is exact, but only some ads carry it.

Can a stopped run continue where it left off?

Yes, for Google ads, with Continue from the last run (continueFromLastRun) on, also on the first run (a run with it off keeps no place). A run that stops before every target is finished (Google refused it, the timeout, your maximum charge or Maximum ads per run) keeps its place in the key-value store google-ads-transparency-continue in your account. The next run with the same input, or the same Monitor name, then:

  • skips the targets an earlier run finished (named in the status message and continueFromLastRun.targetsFinishedEarlier);
  • continues a target it stopped in from the page reached, with the ads already saved counted toward that advertiser's maximum;
  • does not save or charge again an ad an earlier run of this job saved (it reads the earlier runs' datasets back, so ads saved after the last saved place, by a run killed mid-target, are skipped too).

When every target is finished, the place is removed and the next run starts from the beginning. "Same input" means the same targets, countries (Germany = DE), dates as typed (30 days stays the same job the next day), format, platforms, result type, political switch and Domains: only the domain's own advertiser accounts; caps, ad text and details may change. With a Monitor name, the name takes the place of the targets and dates, so you can edit the list; the result type, countries, political switch and own-accounts option still have to match. In advertiser summary mode, an account whose row an earlier run saved is not saved again. Two runs of the same job at once merge what they record.

Limits: a place older than 7 days is not continued (the run starts over and replaces it). Google lists the most recently shown ads first, so its list moves between runs: continuing from a page can miss ads that moved above it meanwhile, and if Google no longer accepts the page reached, the target is read from its start (ads already saved are skipped, not charged). A new-ads watch and runs with several countries read a continued target from its start; their memory or the saved ads skip what was returned. A deleted earlier dataset cannot be read back; the place's own list of saved ads still applies. One place holds up to 20 runs and 100,000 saved ads and forgets neither: a job at 20 runs stops before reading anything, and a run that fills it to 100,000 ads stops there (stoppedBy continue_full, the rest under targetsLeft); start a new job (another monitor name or input) or turn the option off. If the page reached gives no ads, the run says so in the log and inputNotes and reads that list from its start. Cost: no extra charge from this Actor. The run's store writes (at the start, after each target, at most every 15 seconds while a target is read; about 20 to 60 for 20 targets) and its reads of earlier datasets are run usage, included in the per-ad price. The stored ad IDs (about 25 bytes each: about 50 KB for 2,000 ads, at most about 2.5 MB) stay in that store in your account until the job is finished, and Apify can bill kept named storage to your account: at its $1.00 per 1,000 GB-hours (pricing page, read 16 November 2026), 2.5 MB kept for a month is under $0.002.

Several countries in one run

Set Countries (regions), for example ["US", "GB", "DE"]. For each target the run reads Google's list once per country, merges the lists (most recently shown first) and saves each ad once, with matchedRegions naming the countries whose list showed it and searchRegion listing all of them (US,GB,DE). Maximum ads per advertiser applies across all the countries together and each country's list is read up to that maximum, so an ad further down one country's list can lack that country in matchedRegions. Each target's summary entry has regions: ads listed per country, and whether each list was read to its end. A target whose country list was cut at that depth is capped when it reached your maximum; below it (ads skipped by dates, duplicates or earlier runs) it is partial with stoppedBy regions_partial and listed under targetsLeft, and with Continue from the last run on the next run reads those countries deeper. Each extra country adds one list request per 100 ads and a 2 to 5 second pause. Political ads, advertiser summaries and Microsoft Advertising read only the first country, and say so.

How does new-ads-only mode remember what it returned?

It stores ad IDs in a key-value store named google-ads-transparency-seen in your own account, per advertiser, country and format. On a watch's first run it returns the current ads up to your cap and remembers up to 10,000 further ads without charge, so later runs return only ads new since the watch started; they read only the ads shown since the previous run, so a daily check of a large advertiser does not page through its whole history. For larger advertisers it tells old from new by the first-shown time. Each watch remembers at most 50,000 ads per advertiser, country and format. The ads it has returned are never forgotten; to make room it forgets only ads it remembered without charging them, oldest first, and tells those apart by their first-shown time. When the ads a watch has returned reach 50,000, it returns no more new ads for that advertiser, so none is returned or charged twice: the target is listed as unfinished with stoppedBy watch_full and a watchFull note in the run summary. Start a new watch (another Monitor name) to continue. Progress is saved during a run, so after an interrupted run the next one does not return the ads already recorded; ads returned after the last save can come again. Avoid running the same watch twice at once: both runs' memory is kept, but an ad that appears during the overlap can be returned by both. Delete that store to start over.

Can a run cost more than I set?

Not for the ads. Apart from each run's start event, you pay only per ad saved, within limits the run applies together. Maximum ads per advertiser is a hard cap for each domain you enter and for each advertiser account with the same filters, however many of your targets (its ID, names or links) lead to it (ads read for a domain count toward the domain); a longer list stops there and is marked capped. If you let one name match several accounts, the cap applies to each. A maximum charge set in Apify's run options stops the run before another ad would exceed it; saved rows are kept, and the run summary sets budgetExhausted and lists what was not finished under unfinishedTargets. With ad text on, it must cover both the ad and its text before that row is saved. Failed targets and companies without ads: Never charged in the pricing section. In a test on 11 November 2026, a run with a cap of 5 on two advertisers that have hundreds of ads saved 10 ads and reported both as capped.

How do I keep a long list from costing more than I expect?

Maximum ads per run (maxAdsPerRun) caps the whole run, across every domain, name and advertiser; the targets it does not reach are listed under unfinishedTargets with stoppedBy max_ads_per_run and are not charged. Maximum ads per advertiser alone multiplies with the length of the list, and a maximum charge works as above. Every run logs its ceiling at the start (maxPossibleChargeUsd). For example, without ad text, on the Free plan ($0.60 per 1,000 ads), a list of 10,000 domains with the default 100 ads per advertiser could cost up to $600 plus the start event; with Maximum ads per run set to 5,000 it costs at most $3.00 plus the start event. With ad text on, add the ad-text add-on (pricing); the ceiling includes it.

A run reports HTTP 429. What should I do?

For Google: proxy choices are under Google connection settings; what a run does on a block, a refusal or a 429 is under If Google asks for verification in Limits. Run unfinished targets again later; with Continue from the last run on, the same input continues exactly the targets left, without charging again for ads already saved. For Microsoft Advertising: Microsoft Advertising ads in Limits; run the advertisers under unfinishedTargets again later.

Is it legal to scrape the Ads Transparency Center?

The Actor collects only what Google publishes for public transparency and uses no login (Apify's overview). Advertiser names are usually companies, but some advertisers are individuals, so their names can be personal data. Ad wording (headline, description, businessName and text read from ad images) and alert messages, which quote headlines and name the advertisers, can contain the names, phone numbers or addresses of advertisers who are individuals. You are responsible for how you use the results and for complying with the laws that apply to you. This is not legal advice.

Can I see a competitor's Google Shopping ads?

Yes, those Google publishes for that advertiser (ads it does not list cannot be returned). Enter the competitor's domain or name, set Google platforms to Shopping, and turn on Include ad text for the product, price and merchant where Google's preview shows them.

Can I download the results as CSV, Excel or Google Sheets?

Yes: export the run's dataset from its Storage tab or the API as JSON, CSV, Excel (XLSX), XML, HTML or RSS, and import the CSV or Excel file into Google Sheets.

Something is wrong or a field is missing. Where do I report it?

Open an issue on the Issues tab of this page.

Start tracking your competitors' Google ads

Open Google Ads Transparency Scraper and click Start on the prefilled example to see 100 of Nike's ads, then swap in your own competitors. Apify's free plan credit covers about 8,300 ads a month without the ad-text add-on.

More from Fetchwright

Microsoft Ads Library Scraper (Bing ads, EU), Google AI Overview Tracker, Telegram Channel Scraper and Google Trends Scraper.

Changelog

This Actor's version history is a separate document: https://apify.com/fetchwright/google-ads-transparency-scraper/changelog.md

Actor input Schema

targets (type: array):

Required, unless you read targets from a dataset (see Chain with other Actors). One per line: a website domain (nike.com), an advertiser name (Nike), an advertiser ID (AR16735076323512287233) or an Ads Transparency Center URL. A URL's own country, format and dates apply to it where you leave those fields unset. Domains give the most complete results. With Microsoft Advertising chosen below: the advertiser's name as Microsoft lists it (Booking.com B.V.) or its numeric Microsoft advertiser ID (4295086664); Microsoft cannot be searched by website, so a domain is reported as not usable for that target.

adNetwork (type: string):

Which ad library to read. Google (the default) reads the Google Ads Transparency Center. Microsoft Advertising reads Microsoft's ad library of Bing ads shown in the European Economic Area: ad text, landing page, images and, with Include variations, countries, impressions and platforms on, first and last shown dates, the impression range, the share per country, targeting types and who paid. Same price per ad.

maxAdsPerTarget (type: integer):

Stop after this many ads for each domain or advertiser, for example 100. You pay per ad saved, so this caps the cost of a run. Default: 100.

region (type: string):

Country name or two-letter code, such as US, GB, Germany, Deutschland or SE (names are read in English and in the country's own language, without regard to accents). Use ANYWHERE for all countries (the default). A value read this way is noted in the run summary (inputNotes); a value that cannot be read stops the run before any ad is read or charged, naming the value and the accepted forms. For several countries in one run, use Countries below. With Microsoft Advertising, use ANYWHERE (all EEA countries) or one country of the European Economic Area such as DE or FR; with any other country the run searches all EEA countries and says so in the run summary.

regions (type: array):

Optional. Several countries in one run, for example US, United Kingdom, Deutschland. Each advertiser's list is read once per country and each ad is saved and charged once, with matchedRegions naming the countries whose list showed it. The maximum ads per advertiser applies across all countries together, so a run costs no more than with one country. When set, it replaces Country above. Google ads only: political ads, advertiser summaries and Microsoft Advertising read the first country given.

maxAdsPerRun (type: integer):

Stop the whole run after this many ads in all, across every domain, name and advertiser, for example 1000. Use it for long lists, where the maximum per advertiser alone multiplies with the number of targets. Targets not reached are listed as unfinished in the run summary and are not charged. Leave empty for no run-wide maximum.

continueFromLastRun (type: boolean):

On: when a run stops before every target is finished (Google refused it, the timeout, the maximum charge or the maximum ads per run), its place is kept, and the next run with the same input (or the same Monitor name) reads only the targets left, continuing a target from the page it reached. Ads an earlier run saved are never saved or charged again. When every target is finished, the place is removed and the next run starts from the beginning. Turn it on for the first run too: a run with it off keeps no place. The place is kept in a key-value store named google-ads-transparency-continue in your account for 7 days (a few writes per run; Apify lists key-value store writes at $0.05 per 1,000 on Free and Starter). Google ads only. Off (default): every run reads its targets from the start.

resultType (type: string):

Ads lists one row per ad, up to your maximum per advertiser. Advertiser summary returns one row per company saying whether it advertises on Google, roughly how many ads it has, which advertiser accounts run them and in which formats. Use the summary to qualify a list of leads cheaply: it is one row per domain.

format (type: string):

Limit results to one ad format, for example Video for YouTube and video ads or Text for Search ads. Default: all formats.

platforms (type: array):

Keep only ads that ran on these Google platforms, for example YouTube or Google Search. Leave empty for all platforms. Works for every country and costs nothing extra.

dateFrom (type: string):

Keep ads that were still running on or after this date, for example 2026-10-01. Also accepted: 01.10.2026, 1 Oct 2026, and a time before today such as 30 days, 2 weeks or 3 months (counted back from the day of the run). A value that cannot be read (for example 11/09/2026, where day and month are unclear) stops the run before any ad is read or charged. Leave empty for no lower bound.

dateTo (type: string):

Keep ads that started on or before this date, for example 2026-10-31. Also accepted: 31.10.2026, 31 Oct 2026, today, or a time before today such as 7 days. A value that cannot be read stops the run before any ad is read or charged. Leave empty for no upper bound.

politicalAds (type: boolean):

Reads Google's political (election) ads instead of all ads, with the spend range Google states (in the region's currency), times shown and, with details on, the age, gender and location targeting the advertiser chose. Needs a country Google's political ads report covers: AR, AU, BR, CL, GB, IL, IN, MX, NZ, TW, US or ZA. Same price per ad. Google's political ad data names funding organisations and can name natural persons; you are responsible for lawful use. Default: off.

onlyNewAds (type: boolean):

For scheduled monitoring, for example a daily run on your competitors. The Actor records which ads it returned for each advertiser, country and format. With one run of a watch at a time, it returns and charges only ads it has not recorded as returned. Two overlapping runs of the same watch, or a run killed before it saved its memory, can return an ad again. Default: off.

monitorName (type: string):

Optional. Give each scheduled watch its own name, for example sportswear-us, so several new-ads-only watches in one account keep separate memories. Leave empty to identify the watch by its target, country and format.

firstRunBehaviour (type: string):

Applies with Only ads not returned in earlier runs. Return current ads returns the advertiser's current ads up to your cap on the first run. Only remember returns nothing on the first run, so the next run starts from new ads only. In both cases older ads are remembered without charge and are never returned as new later.

notifyWebhookUrl (type: string):

Optional. Paste a Slack or Discord incoming-webhook address, or any https:// address that accepts JSON. When the run saves ads, a short message is posted there: how many ads per advertiser, the first three with their headline and link, and a link to all rows. With Only new ads on a schedule this is a competitor-ad alert. No extra charge.

writeReport (type: boolean):

At the end of the run, turn the ads saved in this run into a report you can send to a client: REPORT.html and REPORT.md in the run storage, linked from the run summary. For example, run three competitor domains and forward the page to the client. Every figure is counted from this run's rows and every sentence comes from a fixed template; no AI model writes it. A report is written when the run has ads from two or more advertisers and at least one advertiser account has its whole list read or at least 100 ads saved. The ads of a domain are often spread over several advertiser accounts, so with domains set "Maximum ads per advertiser" to 200 or more (500 reads most whole lists). Free.

includeAdText (type: boolean):

Adds the ad's wording. For Search text ads it is Google's exact text where Google's preview carries it (adTextStatus found); otherwise the headline, description and display URL are read from the ad image (read_from_image, with a confidence score). In a test on monday.com in the US on 11 November 2026, wording was returned for 39 of 40 ads. For video and app ads it adds the YouTube video, description, call to action and destination URL where Google's preview carries them, and for Shopping ads the product, price and merchant. Each ad also lists the offers its wording states under offers (free trial, discount, promo code, price and more), and with Only ads not returned in earlier runs the run summary says which offers are newly observed since the previous run. Ad text +$0.15 per 1,000 ads with text, on every plan, charged once per saved ad with non-empty wording successfully read. No ad-text charge when the wording is unavailable or this switch is off (the default). One extra request per ad (about 1.3 seconds each); Search text ads read from the image take about 3 seconds each. Give the run 1 GB of memory or more to speed it up.

adTextLimit (type: integer):

With Include ad text on, read the wording for at most this many ads per advertiser, starting with the ads Google lists first, and save the rest without it. Search text ads take about 3 seconds each, so a limit keeps large runs fast: for example 25 gives a wording sample of each advertiser while the full list is still saved. Leave empty to read the wording of every ad.

includeDetails (type: boolean):

Adds every variation of each ad, the countries where it was shown with first and last shown dates, the impression range Google publishes (mainly for ads shown in the EU) the Google platforms it ran on, and which audience criteria the advertiser used where Google publishes them. Makes one extra request per ad, so runs are slower. With Microsoft Advertising: first and last shown dates, the impression range, the share of impressions per country, the targeting types used and who paid, one extra request per ad. Microsoft runs with details are slower: allow about 20 seconds per Microsoft request before source response time and any slowdown; this is not a guaranteed quota.

includeAdvertiserIdentity (type: boolean):

Adds the advertiser's registered country, its verification status (verified true, not verified false, unknown null) and, for verified advertisers, its legal name (advertiserLegalName) to ad rows and advertiser summary rows. Google: an advertiser found by its name gets these from that name search itself, with no extra lookup; for other targets (domains, IDs, links) they come from extra lookups unless already found in an earlier search (under a domain, for the domain's own advertiser accounts, or with none matched its most frequent account: other advertisers' rows keep null values unless already found), and are unknown beyond Advertiser identity lookups per run or when Google's advertiser search does not return the advertiser. Advertiser summary rows of a target given by name carry them from the name search whatever this setting, and with this off, Google ad rows of a target given by name still carry the country from the name search. Microsoft rows take the country and verification status from Microsoft (no legal name). The run summary counts rows with and without identity under identity. No extra charge.

identityLookupLimit (type: integer):

How many extra Google advertiser searches one run may make to find the country and verification status of advertisers not found by a name search (targets given as a domain, ID or link), unless already found in an earlier search. An advertiser found by its name needs no extra lookup and does not count. Advertisers beyond this number keep null values, and the run summary says so under identity.stoppedBecause. Microsoft rows are not affected. Under a domain, lookups go to the domain's own advertiser accounts (isDomainAdvertiser true; with none matched, to the account with the most ads listed), never to an account known to be another advertiser's, so other advertisers linked to the domain do not use them up; their count is under identity.otherAdvertisersNotLookedUp. To get identity for another advertiser, give its advertiser ID as a target. No extra charge; each lookup adds a second or two to the run.

onlyDomainAdvertisers (type: boolean):

Google lists, under a domain, ads of the domain owner and ads of other advertisers it links to that domain (for hubspot.com in the US on 16 Nov 2026, 58 of 200 ads came from 42 other advertisers, mostly HubSpot customers). Every ad row of a domain target carries isDomainAdvertiser: true for the domain's own accounts, false for another advertiser, null when not known. On: only rows with true are saved and charged; the others are left out and not charged, and the run summary counts them. An account is the domain's own when its whole name, without legal endings, spells the domain's name ("Hubspot, Inc." for hubspot.com, not "HubSpot Partner Agency") or its ads show the domain itself or www. of it as their address and it carries a real share of them (at least 3 ads in the domain's list, or 10% of the ads of the accounts matched to the domain); a one-off account linking to the domain is null with linksToDomain true, left out and not charged; an ad on another site marks another advertiser. Addresses come from ad text, so with ad text off only the name decides; wording read only to decide an ad that is then left out does not count toward the ad text limit. A brand whose account name differs from its domain and whose ads carry no readable address can be left out entirely; check domainAdvertisers per target in the run summary. Off (default): every row is saved and charged as before, with the field. Identity lookups go to the domain's own accounts (with none matched, to its most frequent account), either way. A watch remembers other advertisers' ads without returning or charging them. Advertiser IDs, names and links are not affected.

includeTargetsWithoutAds (type: boolean):

In advertiser summary mode, also saves a row for each company with no Google ads in the chosen country (hasAds false, targetStatus no_ads_in_region or no_advertiser_found), so the dataset lines up with your list; each such row is charged as one row. Off by default: a company without ads then gets no row and no charge, and is listed under noAdsTargets in the run summary. A company that could not be checked is never charged and never shown as having no ads: it is listed under failedTargets or unfinishedTargets.

notifyWhenNothingFound (type: boolean):

Also post a message when the run saved no ads. Off by default, so a quiet day sends nothing.

maxAdvertisersPerName (type: integer):

A name such as Nike can match several advertiser accounts. This sets how many to include, starting with the accounts that have the most ads. Default: 1.

targetsDatasetId (type: string):

Optional. Pick a dataset from another run, for example a list of company domains or this Actor's own advertiser summary, and its values are added to the targets above. Use it to chain Actors without exporting anything.

targetsDatasetField (type: string):

Name of the field that holds the domains, names or advertiser IDs, for example domain or website. Leave empty to detect it automatically (domain, website, url, searchTarget, advertiserId, advertiserName, name).

proxyConfiguration (type: object):

For Google runs on Apify, the default is an Apify datacenter session. Omit this field to use it; set {"useApifyProxy": false} to go direct. Your explicit proxy settings are respected. Normal Google reads keep their existing pace, with no per-session request cap. At most two sessions are used in a run. If the first Google request on a datacenter session is blocked or rate-limited, the run waits 60โ€“120 seconds and retries it once on a new session when the timeout allows, with no extra charge. Later refusals stop further Google reads. Locally, omitted settings stay direct. Microsoft uses only the proxy settings you explicitly select.

Actor input object example

{
  "targets": [
    "nike.com"
  ],
  "adNetwork": "google",
  "maxAdsPerTarget": 100,
  "region": "US",
  "continueFromLastRun": false,
  "resultType": "ads",
  "format": "ALL",
  "politicalAds": false,
  "onlyNewAds": false,
  "firstRunBehaviour": "return_current",
  "writeReport": false,
  "includeAdText": false,
  "includeDetails": false,
  "includeAdvertiserIdentity": true,
  "identityLookupLimit": 25,
  "onlyDomainAdvertisers": false,
  "includeTargetsWithoutAds": false,
  "notifyWhenNothingFound": false,
  "maxAdvertisersPerName": 1
}

Actor output Schema

ads (type: string):

One item per ad: advertiser, format, first and last shown dates, preview and Transparency Center link.

summary (type: string):

The run's outcome (complete, partial or failed), ads saved per target, failed and unfinished targets, request count, and whether the spending limit or the timeout stopped the run.

report (type: string):

REPORT.html: a client-ready report of this run, written when "Write a client-ready report" is on. The run summary says when no report was written and why.

API

You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup.

JavaScript example

import { ApifyClient } from 'apify-client';

// Initialize the ApifyClient with your Apify API token
// Replace the '<YOUR_API_TOKEN>' with your token
const client = new ApifyClient({
    token: '<YOUR_API_TOKEN>',
});

// Prepare Actor input
const input = {
    "targets": [
        "nike.com"
    ],
    "maxAdsPerTarget": 100,
    "region": "US"
};

// Run the Actor and wait for it to finish
const run = await client.actor("fetchwright/google-ads-transparency-scraper").call(input);

// Fetch and print Actor results from the run's dataset (if any)
console.log('Results from dataset');
console.log(`๐Ÿ’พ Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`);
const { items } = await client.dataset(run.defaultDatasetId).listItems();
items.forEach((item) => {
    console.dir(item);
});

// ๐Ÿ“š Want to learn more ๐Ÿ“–? Go to โ†’ https://docs.apify.com/api/client/js/docs

Python example

from apify_client import ApifyClient

# Initialize the ApifyClient with your Apify API token
# Replace '<YOUR_API_TOKEN>' with your token.
client = ApifyClient("<YOUR_API_TOKEN>")

# Prepare the Actor input
run_input = {
    "targets": ["nike.com"],
    "maxAdsPerTarget": 100,
    "region": "US",
}

# Run the Actor and wait for it to finish
run = client.actor("fetchwright/google-ads-transparency-scraper").call(run_input=run_input)

# Fetch and print Actor results from the run's dataset (if there are any)
print(f"๐Ÿ’พ Check your data here: https://console.apify.com/storage/datasets/{run.default_dataset_id}")
for item in client.dataset(run.default_dataset_id).iterate_items():
    print(item)

# ๐Ÿ“š Want to learn more ๐Ÿ“–? Go to โ†’ https://docs.apify.com/api/client/python/docs/quick-start

CLI example

echo '{
  "targets": [
    "nike.com"
  ],
  "maxAdsPerTarget": 100,
  "region": "US"
}' |
apify call fetchwright/google-ads-transparency-scraper --silent --output-dataset

MCP server setup

{
    "mcpServers": {
        "apify": {
            "type": "http",
            "url": "https://mcp.apify.com/?tools=fetch-actor-details,fetchwright/google-ads-transparency-scraper"
        }
    }
}

The hosted server signs you in with OAuth on first connect, so no API token belongs in this config. Clients without OAuth support can send an Authorization: Bearer <APIFY_API_TOKEN> header instead, using a token from API & Integrations in Apify Console (https://console.apify.com/settings/integrations).

OpenAPI specification

Download the OpenAPI definition: https://api.apify.com/v2/actors/elL3ELcvHeJ9ASB5g/builds/QSp9c62UfJx7PKFhG/openapi.json