Google Ads Transparency Center scraper by domain, name or ID: dates, impressions, video and new-ad alerts. Base $0.14 per 1,000 ads on Free, $0.11-0.13 on paid plans, usage included. Optional ad text +$0.15 per 1,000 ads with text on every plan. Per-advertiser coverage in each run.
README: a short note and Apify's video on running it on a schedule.
README: related actors updated (same-topic links first, current names).
2026-11-15
README links to runnable examples on GitHub.
Run summary: each advertiser or domain read now carries Google's approximate total for the search as sent (googleAdCountMin, googleAdCountMax, filters applied) next to the ads saved, so you can see how much a cap left unread.
FAQ: "Is there a Google Ads Transparency Center API?" Google's own access is a BigQuery public dataset covering ads shown in the EEA and Turkey; this Actor reads the Transparency Center for the region you choose and works as an API.
0.1.127 - 2026-11-15
A redirect from Google on the first request of a run now gets the same single delayed retry on a new datacenter session as a first-request block; later redirects still end the run quietly.
README opens with a first check: the example task to start from, what it costs (150 ads with ad text: about $0.044 on Free, $0.039 on Business) and one real saved ad. The note about the default datacenter session is in the "Google session reliability" entry below.
Microsoft ad details are included for runs of up to 25 Microsoft ads (counted after packed IDs and name matches are expanded); larger Microsoft runs skip details for advertisers' ads (not charged); single-ad lookups always include details and run for the first 25 only, with or without the details switch, with a note in the log and summary pointing to the Microsoft Ads Library actor, which includes details at any size.
2026-11-15 — Google session reliability
Google runs on Apify default to one pinned Apify datacenter session when proxyConfiguration is omitted. Set useApifyProxy to false for direct access. Explicit proxy settings are respected; omitted local settings stay direct. Previews, images and ad-text fetches keep their existing route.
An eligible verification page or HTTP 429 on the first Google request gets one retry on a new datacenter session after a 60–120 second wait, within the run deadline. At most two sessions; a second block, HTTP 403 or a later refusal stops further Google reads. There is no extra charge for the retry; recovered rows are saved and charged once.
Google reads retain their existing pace and have no per-session request cap. SUMMARY records route, sessionsUsed, firstRequestRetried and the final refusal reason. No automatic residential fallback.
0.1.124 - 2026-11-15
Microsoft runs with details are slower: allow about 20 seconds per Microsoft request before source response time and any slowdown. After an eligible ordinary rate-limit pause, subsequent reads use longer minimum intervals; refusal handling is unchanged. This pacing does not guarantee a quota or completion.
Microsoft ads: a run that sets a maximum charge no longer stops one ad early after a dataset write whose answer was lost; a run never exceeds its maximum charge.
0.1.123 - 2026-11-14
Optional ad text is a paid add-on: +$0.15 per 1,000 ads with text, on every plan. Charged once per saved ad with non-empty wording successfully read, with Include ad text on; no ad-text charge for ads without read wording or with the switch off. The maximum charge covers the ad and its text together, and the pre-run cost ceiling includes the add-on when enabled. Details and the client report remain included.
Price per ad by Apify plan (owner decision 14 November 2026): $0.14 per 1,000 ads on the Free plan, $0.13 on Starter (Bronze), $0.12 on Scale (Silver) and $0.11 on Business (Gold) and above; paid plans are cheaper than Free, and the start event is unchanged. The README price table and cost examples state the price for each plan, and the listing, FAQ and task page the Free-plan price and the paid-plan range, and the price comparison now lists the other actors' prices plan by plan, read on 14 November 2026. The README image no longer shows a price, and the README opening speaks of your competitors' Google ads. The base price comparison also includes automation_craft (21 monthly users), maximedupre (20) and t2solutions (12).
0.1.122 - 2026-11-13
README: the line that the Actor is not affiliated with, endorsed by or sponsored by Google moved from the end of the page to the first section. No change to the Actor, its input or its output.
0.1.121 - 2026-11-13
README: a "Use it in a pipeline" section with one step-by-step recipe for n8n (new ads to Slack), using Apify's own integration and your own Apify account. It says what it was checked against and that it was not run in a live account. New FAQ answers: seeing a competitor's Google Shopping ads, and downloading the results as CSV, Excel or Google Sheets. No change to the Actor, its input or its output.
0.1.120 - 2026-11-13
Reliability: the impressions range of a Microsoft ad is read in one pass, so a malformed range in Microsoft's answer can no longer slow a run down. Results are unchanged.
0.1.119 - 2026-11-13
Reliability: the duration of a video ad is read from its preview in one pass, so a preview that names the video player many times can no longer slow a run down. Results are unchanged.
0.1.118 - 2026-11-13
Reliability: reading sponsored ads from a Google results page no longer slows down on markup with many unclosed tags or comments, which could take seconds before, and an ad whose script element closed without ">" could stop the run from finishing. The ads found are the same as before, except on such broken markup, where the old reading did not finish.
0.1.117 - 2026-11-13
Field description of advertiserCountry: ad rows of a target given by name keep the country from the name search whatever the identity setting (as advertiser summary rows do).
0.1.116 - 2026-11-12
Documentation: the README, input help and field descriptions now state that political ads are read for the 12 countries of Google's political ads report and are not offered for Microsoft, that ads beyond the ad text limit carry no adTextStatus, that a Microsoft page saying the request is blocked ends the Microsoft requests also under HTTP 429, that the identity lookup also adds advertiserLegalName and fills advertiser summary rows, that an advertiser found by its name gets its identity from that name search without counting against the lookup limit, that Microsoft rows take it from Microsoft, the no_wording status, and where advertiserCountry comes from.
0.1.115 - 2026-11-12
A target written for Microsoft Advertising, such as microsoft:Booking.com B.V. or bing:booking.com, in a run set to Google now gets a hint when Google finds no advertiser: the target's note in the run summary and the run's status message say to set Ad network to Microsoft Advertising (EEA) (adNetwork"microsoft") and to enter the name without the prefix. Google is searched as before.
0.1.114 - 2026-11-12
Microsoft Advertising: when Microsoft answers with a page saying the request is blocked, the run asks Microsoft nothing more, also when that page comes with HTTP 429 (before, a 429 with that page was paused out and asked again). What was not read is listed as unfinished and not charged.
0.1.113 - 2026-11-12
New: Political ads only. For the 12 countries Google's political ads report covers, a run returns political (election) ads with the spend range Google states, times shown, the regions they are listed under and, with details on, the age, gender and location targeting the advertiser chose, at the same price per ad. Another actor's topic: "POLITICAL" or political: true, which used to stop the run, now turns it on.
0.1.112 - 2026-11-12
With details on, an ad whose "Times shown" figures Google has not published yet now says from which day they will be (timesShownAvailableFrom). The README now explains audience selection with Google's own caveat: it shows the criteria the advertiser chose, not who saw the ad.
0.1.111 - 2026-11-12
README: Microsoft does state how many ads an advertiser has (up to 1,000); the list of what is not available for Microsoft ads now says which items Microsoft does not publish and which this Actor does not offer for them.
0.1.110 - 2026-11-12
With details on, each Google ad now says who funded it (fundedBy, Google's "Ad funded by"), which can be an agency or another company than the advertiser account, at no extra charge.
0.1.109 - 2026-11-12
Maintenance build: shared code updated. No change in behaviour for this Actor.
0.1.108 - 2026-11-12
Ad rows and advertiser summary rows now include the advertiser's legal name (advertiserLegalName) for advertisers that verified their identity with Google, at no extra charge. Advertiser summary rows also show each account's country and verification status.
0.1.107 - 2026-11-12
README: what a real run with ad wording returned (30 ads, how many had wording and from where), with one example row as saved.
0.1.106 - 2026-11-12
A new input, Maximum ads per run, caps a whole run across every advertiser, so a long list cannot cost more than you set: the run stops saving at that total, and the targets it did not reach are listed as unfinished and not charged. It works for ads, advertiser summaries and Microsoft Advertising. Another actor's run-wide limit (maxItems, maxResults) is read as this cap too.
Every run now states at its start the most it can cost, a ceiling worked out from your limits and the current price, in the log and in the run summary (maxPossibleChargeUsd).
0.1.105 - 2026-11-12
A row is never saved or charged twice when the answer to saving it is lost: the run checks its own dataset, counts a row it finds once (and against your maximum charge), and stops at a row it cannot confirm, listing what it did not finish. The run summary describes it under uncertainWrite.
A new-ads watch no longer forgets an ad it has returned, so such an ad is never returned or charged again. When a watch has returned 50,000 ads for one advertiser, it returns no more new ads for it and says so in the run summary.
0.1.104 - 2026-11-12
If the platform restarts a run (for example to move it to another server), the run now asks for nothing more and saves nothing more, so no ad is saved or charged twice. Its status message and summary say how many rows it saved before the restart; start it again for the rest.
When Google answers with a verification, CAPTCHA or block page (under any status) or a refusal, from any of its data, ad preview or ad image requests, no further request goes to Google in that run, ad details, previews and ad images included. Ads already read are saved without the missing details or wording, the advertisers not read are listed as unfinished, and the summary says why. An ordinary rate limit is handled as before, and ad wording or advertiser names that mention such words are read as data, never taken for a refusal.
An advertiser account reached by two targets with the same filters (its ID and a name, or two names) is now read once, in both result types, so the maximum ads per advertiser holds for the account and an advertiser summary is saved and charged once. The second target is listed under sameAdvertiserTargets and nothing more is charged for it.
An ad with an unreadable date no longer stops the rest of its advertiser's list; the date is left empty.
README: what ad wording and alerts can contain about individual advertisers, and the cap and verification sentences now match the behaviour.
When a run stops because Google asked for verification, a short record of what Google answered (status, type, length and the verification phrases, not the page) is kept as REFUSAL_PAGE in the run's key-value store.
0.1.103 - 2026-11-12
README: the "only new ads" sentences now state their condition and exceptions (one run of a watch at a time; a run killed before it saved its memory can return an ad again), the cost of a 1,000-domain check includes the start fee, a target that fails part-way is described as charged for the ads saved before it failed, and the two rival price ranges now say which actors and dates they cover.
The run's status message no longer says a failed target was not charged when ads were saved for it before it failed; it says such a target is charged only for those ads.
0.1.102 - 2026-11-12
README and input help now say plainly what new-ads-only mode guarantees: with one run of a watch at a time, the ads it has recorded as returned are skipped and not charged; two overlapping runs of the same watch, or a run killed before it saved its memory, can return an ad again.
0.1.101 - 2026-11-12
README: the Python example now matches the current Apify Python client.
0.1.100 - 2026-11-11
README: a new section, "Coming from another Google Ads scraper?", lists how each field name from another actor's saved input is read, not applied or stops the run, the defaults that differ, and prices; a test keeps its table in line with the code. No change in behaviour.
0.1.99 - 2026-11-11
If Google answers a run with a verification page or a refusal instead of data, the run now stops asking Google at once and does nothing to get past it. The advertisers not read are reported with the reason, and only saved rows are charged.
0.1.98 - 2026-11-11
Offers: price guarantees are now found as guarantee with value best price, lowest price or price ("Best price guaranteed", "Mejor precio garantizado", "Precio mínimo garantizado", "Bestpreisgarantie", "Preisgarantie", "Meilleur prix garanti", "Laagste prijs gegarandeerd", "Miglior prezzo garantito").
Offers: a new type, free_cancellation ("Free cancellation", "Kostenlose Stornierung", "kostenlos stornierbar", "Cancelación gratis", "Annulation gratuite", "Gratis annuleren", "Cancellazione gratuita").
Offers: a price cut stated without a figure ("Reduced prices", "Prix cassés", "Precios rebajados", "Reduzierte Preise", "Prijzen verlaagd", "Prezzi ridotti") is a discount with value null. "Deals", "Angebote" or "save" on their own are still not offers, and a satisfaction figure such as "98% Kundenzufriedenheit" is not a discount.
These are found in English, Spanish, German, French and Dutch, and in Italian, by fixed word rules, so an offer worded another way is still missed; a negated one ("Keine Preisgarantie", "No free cancellation") is not listed. Offers found before are unchanged.
0.1.97 - 2026-11-11
A pasted Transparency Center link now keeps its own country, format and dates (region, format, start-date, end-date) for that link's advertiser or domain, where the input leaves those fields unset. Before, such a link was read with the input's own filters only, so a run whose input set none returned every country, format and date. A country, format or date set in the input still applies to every target, and the run notes where a link asked for something else. Rows carry the target's own country in searchRegion.
A link to one ad still returns that ad: its country is used, its format and dates are not, and the run says so. In advertiser summary mode only a link's country is used.
The same advertiser pasted with different filters is read once per set of filters, each up to the maximum ads per advertiser; the run says so, and an ad found twice is saved and charged once.
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, so ads it already returned are not returned or charged again.
More field names from inputs saved for other Ads Transparency actors are read: searchTerms, maxAds, shownAfter and shownBefore, adFormats with one format, includeRegionStats and monitorStoreName.
adFormats with two or more formats, and a topic other than all (in the input or in a link), now stop the run before any ad is read, since this Actor cannot apply them and would charge for ads the input excludes; nothing is charged apart from the start event, and the message says what to set.
Fields that would raise the cost (advertisers per name, "no limit") or that this Actor does not offer (media downloads, cookies, link lists from a file) are not applied, and the run says what to set instead.
The run summary has a new inputNotes list with how the input was read: other field names, fields not applied, and filters taken from links.
Microsoft Advertising rows now link to the ad's page in Microsoft's Ad Library instead of a data address.
0.1.96 - 2026-11-11
Microsoft Advertising runs close to their timeout now stop asking Microsoft in time to save the run summary, instead of waiting out a pause first. No change for Google runs.
0.1.95 - 2026-11-11
The listing description leads with Google Ads Transparency Center again; Microsoft's ad library now also has its own Actor, Microsoft Ads Library Scraper. No change in behaviour.
0.1.94 - 2026-11-11
Microsoft Advertising runs: status messages now say "no ad charge" where nothing was read, so they no longer read as if the start event were waived, and a run that stops on Microsoft's rate limit says why (limit kept on after two pauses, a wait longer than two minutes was asked, or too little run time was left).
0.1.93 - 2026-11-11
Microsoft Advertising runs now keep a steadier pace from the start (about 180 ads a minute without details), so a run no longer loses a minute to a pause in its first seconds. In our test on 11 November 2026 a 500-ad run met no limit at this pace.
0.1.92 - 2026-11-11
Microsoft Advertising runs: when Microsoft limits how often it is asked (HTTP 429), the run now pauses, slows down and continues instead of stopping. A refusal (HTTP 403 or Microsoft's "The request is blocked" page), or limiting that goes on after two pauses, ends the Microsoft part of the run; the advertisers not finished are listed as unfinished and not charged for what was not read.
0.1.91 - 2026-11-11
Microsoft Advertising runs: when Microsoft refuses a request (HTTP 403 or 429, or its "The request is blocked" page), the run now stops asking Microsoft at once instead of waiting and trying again; the advertisers not reached are listed as unfinished and not charged. A run with a proxy now uses one proxy address throughout.
Microsoft Advertising runs: an advertiser entered twice (by name and by its numeric ID) is now read once, so Maximum ads per advertiser holds for it; the second entry is listed under sameAdvertiserTargets in the run summary.
Microsoft Advertising runs: an advertiser stopped by an error after some of its ads were saved is now listed under unfinishedTargets, and the status message says the saved ads were charged and the rest was not, instead of calling the advertiser not charged.
Microsoft Advertising runs with a country chosen now wait at most 60 seconds for Microsoft's ad list instead of about 150 seconds, and the message says to set the country to ANYWHERE. Requests to Microsoft also end before the run's timeout, so the summary is always saved.
Microsoft Advertising rows whose detail lookup came back empty now say so in detailsError instead of looking like rows with details.
Offers: a negated offer such as "No free trial" or "Kein kostenloser Versand" is no longer listed. For Microsoft Advertising runs, the offers note in the run summary no longer points to a new-ads option those runs do not have, and says offers are found by word rules for English, German, French, Spanish and Dutch.
0.1.90 - 2026-11-11
Microsoft Advertising runs no longer stop when the country is outside the EEA, which is what the form's suggested US is. The run searches ads shown in any EEA country and states that in the log and in regionNote of the run summary.
0.1.89 - 2026-11-11
A target given twice under two field names of another actor (for example in both domains and queries) is now looked up once.
0.1.88 - 2026-11-11
Fix for 0.1.87: a limit or format given under another actor's field name (maxAdsPerSearch, resultsLimit, maxResults, adFormat) is now honoured. In 0.1.87 only the targets were read, so such a run used the default maximum of 100 ads.
0.1.87 - 2026-11-11
Input written for other Google Ads Transparency actors now works as it is: field names such as domains, queries, advertiserIds, startUrls, searchQuery, maxAdsPerSearch, resultsLimit, maxResults, startDate, endDate, adFormat and platform are read as this Actor's own, and the log says how they were read. A field given under this Actor's own name always wins.
0.1.86 - 2026-11-11
The run's storage now describes its records (the run summary, the client report), so they show as named tabs in the run's Storage view, and the main input's help says it is required unless a dataset is used. The details option now sits with the other per-ad options in the input form. No change to results or prices.
0.1.85 - 2026-11-11
The client report's help now gives the full condition: ads from two or more advertisers, and at least one advertiser account with its whole list read or 100 ads saved. (0.1.84 left out the first part.)
0.1.84 - 2026-11-11
The help for the client report now states the real condition (one advertiser account with its whole list read or 100 ads saved) and the run warns at the start when the maximum is likely too low for a report on a domain. No change to results or prices.
0.1.83 - 2026-11-11
The client report (Write a client-ready report) now has an Offers section when Include ad text is on: for each advertiser, the offers its ads state (for example "Free trial, 14 days" or "Discount 20%"), how many of its ads with read wording state each one, and one ad as proof with its link. With Only ads not returned in earlier runs, it also lists the offers newly observed since the last run and, when the run summary has them, those no longer observed; otherwise it says why in one sentence, and on a watch's first run it says that run is the starting point. A newly observed offer is one this watch had not seen; it may have run before in ads whose wording was not read. Reports from runs without ad text are unchanged.
0.1.82 - 2026-11-11
With Include ad text on, each ad now has offers: the free trials, discounts, promo codes, free shipping, demos, webinars, free plans, prices, guarantees and deadlines its wording states, each with its value and the exact phrase. English, plus common phrases in German, French, Spanish and Dutch. Offers have no separate charge; since 0.1.123, ad text adds $0.15 per 1,000 ads with text on any plan.
The run summary has an offers object: per advertiser, the offers observed in this run with one ad as proof. With Only ads not returned in earlier runs, it also lists the offers newly observed since the watch's previous run, and those no longer observed when the advertiser's list was read to the end and the wording of its recently shown ads was read. The first run of a watch is a baseline. Up to three newly observed offers are added to the alert.
Microsoft Advertising rows carry offers too.
0.1.81 - 2026-11-11
The listing now states plainly what bounds a run's cost (the cap per advertiser, your maximum charge, no charge for failed or empty targets) and no longer suggests that a long-running ad is a proven performer. No change to results or prices.
0.1.80 - 2026-11-11
New Ad network option: choose Microsoft Advertising to get the Bing ads Microsoft publishes for ads shown in the European Economic Area, by advertiser name or Microsoft advertiser ID, at the same price per ad. Rows carry the headline, description, display and landing URL and images, and with details on the first and last shown dates, the impression range, the share of impressions per country, the targeting types used and who paid. Google remains the default and nothing changes for Google runs, except that Google rows now carry adNetworkgoogle.
The run summary of a Microsoft run counts failures by reason (failuresByReason) and lists companies without ads or without an advertiser under noAdsTargets, not charged.
0.1.79 - 2026-11-11
The run summary now says how much of each advertiser's list was read, separately from whether the run did what you asked: coverage is complete, capped (stopped at your maximum), partial (stopped by the maximum charge, the timeout or an error) or unknown, for each target and for the run. Targets that were never started now appear in targets too.
Alerts to Slack, Discord or a webhook are tried up to three times when the receiver is busy or briefly failing. The run summary's notification always says what happened (sent, failed, not_configured or nothing_to_send, with the HTTP status, error and number of attempts), and the status message says when an alert could not be delivered. Saved rows are never affected.
The run summary's new identity object counts the rows with and without the advertiser's country and verification and says when the lookup limit was reached. A new option, Advertiser identity lookups per run (default 25), lets you fill it for more advertisers.
Clearer wording on the listing: what a run covers, which companies without ads are charged (only rows you ask for with also save companies without ads), and which Search text ads come with Google's exact text.
0.1.78 - 2026-11-11
The listing now has copy-and-paste examples for Python, Node.js and curl. No change to results or prices.
0.1.77 - 2026-11-11
Search text ads: one more form of the advertiser icon is handled, and a headline that itself looks like a web address ("Try monday.com") is no longer taken for the ad's display URL.
0.1.76 - 2026-11-11
More Search text ads return their wording. When the advertiser's icon in the ad image was read as a letter, the ad came back as image_unreadable; in a test on monday.com this was 6 of 20 text ads. They are now read.
0.1.75 - 2026-11-11
An ad image that does not download is now fetched up to three times before the ad is marked image_unreadable. The run summary says why images were not read (adText.imageFailures) and now counts ads that have a preview (adText.withPreview, which always showed 0).
0.1.74 - 2026-11-11
When an ad's preview does not load, it is now tried a third time after a longer wait, so fewer ads come back as preview_unavailable. The run summary counts the previews that still failed, by reason, under adText.previewFailures.
0.1.73 - 2026-11-11
Search text ads that Google publishes with a preview now return their exact headline, description, display URL and sitelinks. Before, these ads came back without wording (not_in_preview).
0.1.72 - 2026-11-11
Maintenance build. No change to results or prices.
0.1.71 - 2026-11-11
The listing now compares features with the four most-used Google Ads Transparency actors, from their public listings, and says where they do more.
0.1.70 - 2026-11-11
A domain, advertiser ID or link pasted with quotes, brackets, a list marker, a port or closing punctuation ("nike.com", - nike.com, nike.com:443, nike.com.) is now read as that advertiser. Before, it was searched as a company name. An advertiser ID typed in lower case works too.
The same advertiser given twice in different forms (nike.com and https://www.nike.com/ ) is read and charged once.
0.1.69 - 2026-11-10
Several domains, advertiser IDs or Transparency Center links pasted on one line ("nike.com, adidas.com") are now read as separate advertisers. Company names with commas or spaces ("Hubspot, Inc.") are left whole. "@nike.com" is read as the domain.
0.1.68 - 2026-11-10
The Output tab's client report link now always opens a page: the report when one was written, otherwise a short note saying why there is none and what to change, with a link to a sample report.
0.1.65 - 2026-11-10
New: Ad text for the first ads only. With ad text on, set a number and the wording is read for that many ads per advertiser, the rest are saved without it. A sample of 25 per advertiser adds about a minute instead of making a large run many times longer. The run summary records the limit.
A long run now reports its progress every few seconds in the status message and the log: the advertiser being read and how many ads are saved so far.
The client report no longer calls an ad an advertiser's oldest when its list was only read in part, says "1 day", and states "all ad formats" when the run had no format filter.
0.1.62 - 2026-11-10
New: a client-ready report, free. Switch on "Write a client-ready report" and the run also saves REPORT.html and REPORT.md: the strongest finding about the advertisers in this run, their launches week by week, real ads as proof, and what the data cannot show. Every figure is counted from the run's rows; no AI model writes it. It never changes the run's results or charges.
0.1.61 - 2026-11-10
A name search with a country now reads the advertiser account with the most ads in that country, not the largest account worldwide. The run summary says how the account was chosen (chosenBy) and shows each other match's ad count in that country.
0.1.57 - 2026-11-10
New: alerts. Paste a Slack or Discord incoming-webhook address (or any HTTPS address) and each run posts a short message with the number of ads per advertiser, the first three with headline and link, and a link to all rows. A quiet run sends nothing unless you ask for it.
0.1.56 - 2026-11-10
Ad text: lone divider marks no longer appear in descriptions. Documentation: the ad text figures now come from a hand-checked sample.
0.1.55 - 2026-11-10
Ad text of Search ads: one more ad layout is now read, "AI" is no longer returned as "Al", and site links and review snippets no longer run into description. Site link titles are returned in the new sitelinks field.
0.1.53 - 2026-11-10
New-ads watches save their progress while a run works, so a run that is interrupted does not return the same ads again on the next run.
Two runs of the same watch at the same time no longer overwrite each other's memory.
0.1.50 - 2026-11-10
When a name matches several advertiser accounts, the run summary now lists the accounts that were not read (otherMatches), so a wrong match is easy to spot.
The run summary says for each advertiser whether its list was read to the end (listReadToEnd) and what stopped it if not.
Advertiser summary mode: new option to save a row for companies without ads too, with a targetStatus field, so the dataset lines up with your list.
Documentation: clearer wording on what the count check and the ad text figures do and do not show.
0.1.49 - 2026-11-10
Every ad row now carries advertiserCountry and advertiserVerified (whether Google shows the advertiser as identity-verified), for domain, name and ID targets alike. On by default at no extra charge; null when Google's advertiser search does not return the advertiser.
Advertiser summary rows from a name search also carry advertiserVerified.
0.1.48 - 2026-11-10
Documentation: a dated price comparison with the five most-used actors in this category.
0.1.47 - 2026-11-10
Scheduled new-ads runs are shorter: after a watch's first run, a run reads only the ads shown since the previous run.
A name now matches the one advertiser account with the most ads by default (it was three), so similarly named accounts are not pulled in. Raise Advertisers per name search to include more.
adTextStatus has a new value no_wording for ads whose preview gave a landing page, video or app but no wording; found now always means wording.
A video is named only when the preview shows exactly one.
A warning when a dataset of targets holds more than 10,000 items.
0.1.46 - 2026-11-10
Documentation: a completeness check against the ad counts Google publishes (four advertisers, 10 November 2026).
0.1.45 - 2026-11-10
New images in the documentation; ad text and details speeds stated as measured.
0.1.44 - 2026-11-10
Ad text is now fetched for several ads at once: in tests on 10 November 2026, 117 video ads with ad text took 41 seconds (74 seconds before). Search text ads, read from the ad image, still take about 3 seconds each.
0.1.43 - 2026-11-10
Details lookups several at a time now hold up when Google limits some of them: in tests on 10 November 2026, 60 ads with details took 49 to 55 seconds (94 to 97 seconds before).
0.1.41 - 2026-11-10
Runs with variations, countries, impressions and platforms switched on are faster: details are now looked up several at a time.
Documentation rewritten for clarity: who it is for, why choose it, use cases by role.
0.1.40 - 2026-11-10
Documentation: measured coverage of ad text and video fields for image and video ads.
0.1.39 - 2026-11-10
With ad text on, product ads shown as images now return the product title and merchant, more ads return their landing page in destinationUrl, and more video ads return their YouTube video.
0.1.38 - 2026-11-10
Documentation: measured coverage of the video fields.
0.1.37 - 2026-11-10
With ad text on, many more video ads now carry youtubeVideoId, videoUrl and videoThumbnailUrl.
Ads for a business location now return their headline, description, call to action and display URL, plus businessName, businessCategory and advertiserDomain.
0.1.36 - 2026-11-10
With ad text on, video ads now carry youtubeVideoId, videoUrl, videoThumbnailUrl and videoDurationSeconds, and app ads appId and appStoreName, where Google's preview names them.
The country field also accepts three-letter country codes.
0.1.35 - 2026-11-10
The country field accepts English country names (for example Germany or United States) as well as two-letter codes.
0.1.34 - 2026-11-10
New-ads-only watches no longer return older ads as new for large advertisers. Older ads are remembered without charge. New option: on a watch's first run, return the current ads or only remember them.
Advertiser summary: companies with no ads are listed in the SUMMARY record and no longer charged.
A single ad's Transparency Center link now returns just that ad. A link that cannot be read is reported as a failed target; the other targets still run.
An ad matching several of your targets is saved and charged once per run.
The SUMMARY record has outcome (complete, partial or failed); the status message of a partial run starts with "Partial:".
matchedPlatforms with several platforms chosen now lists only platforms confirmed for the ad.
0.1.33 - 2026-11-10
Documentation: the connector address for using this Actor as a tool in Claude, ChatGPT and other AI apps.
0.1.32 - 2026-11-10
Long runs now finish cleanly before the run timeout: every saved row is kept, the run succeeds, and targets that were not completed are listed under unfinishedTargets in the SUMMARY record so you can rerun only those.
The country field accepts UK as well as GB.
New images in the documentation.
0.1.31 - 2026-11-10
Maintenance release. No change in behaviour.
0.1.27 - 2026-11-10
The Google platforms filter now works for every country and is much faster. Rows carry matchedPlatforms.
0.1.24 - 2026-11-10
Reliability: runs are more resilient when Google rate-limits requests.
0.1.21 - 2026-11-10
New: Google platforms filter (Search, YouTube, Shopping, Maps, Play).
New: Read targets from a dataset, to chain this Actor after another one.
The final status message now says what was skipped and what was not charged.
0.1.19 - 2026-11-10
Two more result views: Ad text and Advertiser summary.
Input form reordered: the ads cap now comes second; clearer field help with examples.
README: Limits section and two more FAQ answers.
0.1.17 - 2026-11-10
Reliability: runs now recover by themselves when Google rate-limits requests.
0.1.16 - 2026-11-10
Price lowered to $0.40 per 1,000 ads on the Free plan, down to $0.30 on Business and above. At this version, ad text, impressions and the advertiser summary were included; since 0.1.123, ad text adds $0.15 per 1,000 ads with text on any plan.
0.1.15 - 2026-11-10
New: Include ad text. Headline, description and display URL of Search text ads are read from the ad image, with a confidence score per row. Video and app ads get description, call to action and destination URL where Google's preview carries them; Shopping ads get product, price and merchant.
Default memory is now 512 MB.
0.1.11 - 2026-11-10
Added this changelog.
0.1.9 - 2026-11-09
New: impression range and Google platforms per country, with first and last shown dates, when details are included.
New: Advertiser summary mode. One row per company: whether it advertises on Google, approximate ad count, advertiser accounts and formats.
New: Monitor name, so several new-ads-only watches in one account keep separate memories.
0.1.6 - 2026-11-09
Lower, tiered prices: $1.50 per 1,000 ads on the Free plan down to $0.80 on Business.
0.1.5 - 2026-11-09
Fixed: a search by company name now picks the advertiser accounts with the most ads first, so the real brand is no longer missed among similarly named advertisers.
Output fields are now described in the dataset schema.
0.1.2 - 2026-11-09
First public release: search by domain, advertiser name, advertiser ID or Transparency Center URL; country, format and date filters; new-ads-only mode; variations and countries.