All notable changes to Google Ads Transparency Scraper. Dates are UTC. Small internal refactors and documentation-only updates are omitted.
- Faster recovery from Google rate limits: shorter, jittered backoff on HTTP 429/403/5xx,
Retry-After honoured (capped at 60 s) and a per-run waiting budget (about 90 s at most), so a throttled run finishes well before its timeout instead of timing out.
- New time budget: the run wraps up about 15% before its timeout and keeps every ad delivered so far. New optional input
maxRunSeconds.
- Targets interrupted by rate limiting or the time limit now get a free
error row (throttled, partial, resultsBeforeStop); targets that could not be checked in time are marked skipped.
- When every target was only rate-limited, the run now ends as SUCCEEDED with free error rows and a clear status message, so n8n, Make and AI agents get an explanation instead of a generic failed run. Real problems (invalid input, HTTP 400, network or proxy errors) still fail the run.
- Only new ads mode: an incomplete target no longer advances its "last run" marker, and ads already delivered are never charged twice.
- Only new ads (
onlyNew, optional stateKey): scheduled runs return only creatives that started running since the previous run. Already-seen and older ads are skipped for free; maxAdsToCheck caps the scan for very large advertisers. Also works in advertisers mode (only new advertiser accounts).
- New fields on every ad row:
isActive, daysSinceLastShown and isNew; previewUrl added to the dataset view.
- Advertiser IDs (
AR...) and Transparency Center links pasted into queries are detected automatically.
- Default region is now anywhere (previously US).
- Brand names now resolve to the right advertiser for the selected countries: region-aware suggestions, exact-name ranking, fallback to the next matching account and then to the brand's website.
- An API call with only
domains or advertiserIds no longer also scrapes the demo brand.
- Invalid input (non-numeric limits, unknown
resultType, start date after end date) stops with a clear message and nothing charged. Numbers and booleans sent as strings are accepted.
- Clearer zero-result messages that list the closest matching advertisers and websites.
- Ad details fetched for a pasted creative link are no longer charged as an add-on unless
includeDetails is on.
- Non-JSON or truncated responses from Google are retried with a fresh session.
- Without a proxy, repeated HTTP 429 responses switch the run to Apify Proxy automatically (unless
useApifyProxy is explicitly false).
- Pagination stops if Google repeats a page token, so a run can never loop.
- The empty-input demo is limited to 20 ads.
- A run started without any target field (for example a bare
{} API call) now runs a small documented demo instead of failing. Targets that are sent but empty stop with a clear message and nothing charged.
- HTTP 403 from Google is handled like a rate limit (new session, backoff), and
Retry-After is honoured.
- Requests rejected by Google (HTTP 400 and other 4xx) stop immediately with a clear message instead of being retried.
- If Apify Proxy refuses the connection twice, the run continues without proxy instead of failing.
- Input is normalised and validated before any request: country names and aliases (
UK to GB, Germany to DE), several date formats, a single string where a list is expected, and internationalised domains.
- Export ads from the Google Ads Transparency Center by brand name, domain, advertiser ID or Transparency Center URL: advertiser identity, format, first/last shown dates, image and preview links.
- Optional ad text and YouTube video IDs, and optional full details (per-country dates, EU reach and audience targeting, variations, topic).
- Advertisers mode: which accounts advertise a brand or website.
- Filters by region, format and date range; pay per delivered ad, empty searches and errors are free.