## Google Ads Transparency Scraper
**π° $1.20 / 1 000 ads** Β· pay only for results Β· no credit card to try
_We do the dirty work so your dataset stays clean._ π
Extract every ad creative from **Google Ads Transparency Center** for a brand domain or advertiser ID. Get advertiser name, creative ID, landing domain, format type, first/last seen timestamps, and preview-image / video URLs β exported to JSON, CSV, or Excel.
---
### π― What this scrapes
The [Google Ads Transparency Center](https://adstransparency.google.com) is Google's public registry of every ad campaign running on Search, YouTube, Display, Shopping, Maps, and Play. This Actor talks to its internal `SearchService/SearchCreatives` RPC directly β no Chrome, no Playwright, no Selenium β so it's fast, stable, and cheap.
Per creative you get:
| Field | Type | Notes |
|---|---|---|
| `advertiser_id` | string | Google's stable advertiser identifier (e.g. `AR0123β¦`) |
| `advertiser_name` | string | Public-facing brand name |
| `creative_id` | string | Stable per-creative ID (e.g. `CR0123β¦`) |
| `creative_url` | string | Deep link into Transparency Center |
| `landing_domain` | string | Click-through domain |
| `format_type` | integer | Numeric format code (1=text, 2=image, 3=video β inferred) |
| `first_shown_ts` | integer | Unix seconds, first observed impression |
| `last_shown_ts` | integer | Unix seconds, last observed impression |
| `impressions` | integer | Google-reported impression count |
| `preview_image_url` | string \| null | Static thumbnail (image creatives) |
| `preview_content_js_url` | string \| null | JS bundle URL (video/rich creatives) |
| `region` | string | Locale label you passed (display only) |
| `scraped_at` | string | ISO-8601 UTC timestamp |
### π‘ Use cases
- **Competitor ad-spend tracking** β pull every Nike ad once a week and diff the creative set to see launches.
- **Trademark enforcement** β monitor advertisers running ads against your brand keyword. Combine with your own takedown workflow.
- **Affiliate-fraud detection** β flag advertisers whose landing domain doesn't match the advertiser name.
- **Political-ad monitoring** β track which advertisers are active in an election cycle.
- **Brand-safety audits** β for agencies, prove the ads currently live for a client.
- **Market research** β observe how saturated a vertical (e.g. crypto, sports betting, supplements) is with ads.
- **AI / RAG ingestion** β feed creative metadata + image URLs into a vector store for image-grounded analysis.
### βοΈ How to use it
1. **Click "Try for free"** at the top of the page.
2. **Paste one or more brand domains** into the `Brand domains` field (e.g. `nike.com`, `adidas.com`). One per line. Each domain spawns its own scrape.
3. *(Optional)* Drop in **advertiser IDs** if you already know them β they look like `AR0123456789` and live in the Transparency Center URL when you click into an advertiser.
4. *(Optional)* Set a **date window** to narrow ad activity. Defaults to the last 365 days.
5. **Run**. Each ad is one row in the dataset; export to JSON, CSV, or Excel from the Storage tab.
The first run on a new account uses `$5` of free Apify credit β that's roughly **4 000 ads** at our pricing.
### π₯ Input
The schema lives in `.actor/input_schema.json`. The fields:
| Field | Type | Required | Default | Notes |
|---|---|:--:|---|---|
| `searchDomains` | array of string | one of | `["nike.com"]` | Brand landing domains, one per line |
| `advertiserIds` | array of string | one of | `[]` | Google advertiser IDs (`ARβ¦`) |
| `region` | enum string | no | `anywhere` | **Display-only** β Google's RPC does not filter by region (see [Limitations](#limitations)) |
| `dateFrom` | string (`YYYY-MM-DD`) | no | 365 days back | Lower bound of ad-activity window |
| `dateTo` | string (`YYYY-MM-DD`) | no | today (UTC) | Upper bound |
| `maxResults` | integer | no | `1000` | Total dataset items across all targets. `0` = unlimited |
| `maxPages` | integer | no | `25` | RPC budget per target (40 ads Γ 25 pages = 1 000 / target) |
| `proxyConfiguration` | proxy config | no | Apify Proxy enabled | Sticky session recommended |
At least one of `searchDomains` or `advertiserIds` must contain at least one entry.
#### Example input
```json
{
"searchDomains": ["nike.com", "adidas.com"],
"advertiserIds": ["AR03012025048987521025"],
"region": "US",
"dateFrom": "2025-11-15",
"dateTo": "2026-05-15",
"maxResults": 5000,
"maxPages": 25,
"proxyConfiguration": { "useApifyProxy": true }
}
````
### π€ Output
Every row is one creative. Example:
```json
{
"advertiser_id": "AR18378488041124659201",
"advertiser_name": "Nike Retail BV",
"creative_id": "CR15771942603307614209",
"creative_url": "https://adstransparency.google.com/advertiser/AR18378488041124659201/creative/CR15771942603307614209?region=anywhere",
"landing_domain": "nike.com",
"format_type": 1,
"first_shown_ts": 1761145807,
"last_shown_ts": 1778871417,
"impressions": 205,
"preview_image_url": "https://tpc.googlesyndication.com/archive/simgad/12774179880874022668",
"preview_content_js_url": null,
"region": "anywhere",
"scraped_at": "2026-05-15T19:17:59+00:00"
}
```
Export options once the run finishes:
- **JSON** β full payload, ideal for AI/RAG pipelines
- **CSV / Excel** β for analyst spreadsheets, sort by `impressions` to find big-spender ads
- **JSONL** β line-delimited, easy to stream into a warehouse
- **API** β fetch programmatically via `GET /v2/datasets/{id}/items`; webhook on `ACTOR.RUN.SUCCEEDED` for live pipelines
### π° Pricing
Pay-per-event. You pay for what you get, nothing for what you ask for:
| Event | Price | When charged |
|---|---:|---|
| `actor-start` | **$0.005** | Once per run (covers warm-up + cookie handshake) |
| `ad-result` | **$0.0012** | Per ad creative written to the dataset |
**Examples**:
| Pull | Cost |
|---|---:|
| 100 ads | $0.13 |
| 1 000 ads | $1.21 |
| 10 000 ads | $12.01 |
| 100 000 ads (monthly competitor sweep) | $120.05 |
Compare to: building this in-house is ~2 engineer-weeks plus the ongoing cost of maintaining a proxy pool and the TLS-fingerprint replay loop. We've already done it.
### π₯ Features
- β‘ **Direct-RPC, no browser** β calls Google's internal `SearchCreatives` endpoint via `curl-cffi` with a Firefox TLS+H2 fingerprint. ~1 second per page (40 ads).
- π **Batch input** β scrape multiple domains and advertiser IDs in a single run, deduplicated automatically.
- π‘οΈ **Apify Proxy aware** β sticky-session URLs threaded through curl-cffi for cookie continuity across pagination.
- π¦ **Golden-file tested** β every parser change runs against 4 captured creative shapes (still image, rich video, minimal, malformed) before shipping.
- π¬ **Live wire validation** β opt-in smoke tests catch RPC contract drift before users do.
- πͺΆ **Open output schema** β flat, predictable JSON. Drops straight into Pandas, BigQuery, or your vector store.
### π‘ Tips for best results
- **Search by domain when you can** β domains return *every* ad pointing at that landing page, including ones bought by resellers or affiliates. Advertiser IDs are narrower.
- **Pull bigger date windows in one run** β Google's response is paginated, not date-segmented. Asking for 365 days is the same cost per ad as asking for 30 days.
- **Big brands hit the cap fast** β Nike's library claims ~300 000 ads. The default `maxPages=25` caps at 1 000 ads per domain. Raise it for full-history pulls.
- **Use `region` as a tag** β it doesn't filter the RPC, but if you're running parallel campaigns per market, set `region: "US"` on the US sweep so your downstream tables can group by it.
- **Run weekly, not hourly** β Google updates the Transparency Center daily at most. Scheduling more often just wastes credit.
### π§ Limitations
- **Region is metadata, not a filter.** Google's `SearchCreatives` RPC ignores the geo target β we proved this empirically (see `scripts/recon/FINDINGS.md`). The Transparency Center's browser UI shows a region selector, but the server returns the same creative set regardless. We expose `region` purely so you can tag exports by intended locale.
- **No region-only browsing.** You must give the scraper a `searchDomain` or `advertiserId`. There is no "all ads in country X" mode on the public RPC. If Google adds one we'll ship it.
- **Video / rich creatives return a `content.js` URL, not an MP4.** Rendering the actual video frame requires executing Google's JS bundle β out of scope for v1.
- **Date range is enforced by Google, not us.** They retain ~12 months of history. Asking for older dates just clips to that window.
### π Integrations
- **Schedule** β Apify Console β Schedules tab β run weekly for monitoring.
- **Webhooks** β register `ACTOR.RUN.SUCCEEDED` to fire your downstream pipeline as soon as the dataset is final.
- **API** β `POST /v2/acts/DevilScrapes~google-ads-transparency/run-sync-get-dataset-items` returns the full result set in one synchronous call (good for β€ a few thousand ads).
- **Make / Zapier** β every Apify Actor surfaces as a node out of the box.
- **n8n** β use the Apify community node.
### β FAQ
**Is this legal?**
Yes. The Google Ads Transparency Center is a public registry Google operates under EU DSA + US regulatory pressure. We scrape only what the public UI exposes, at a polite cadence (~1 req/s per session), and we do not bypass any authentication. We also don't collect personal data β only advertiser-level metadata.
**Why is the region selector marked "display only"?**
Because we empirically confirmed the RPC ignores it. Other scrapers on the Store claim region filtering; we tested every plausible RPC body shape and none of them returned a region-narrowed result set. We'd rather under-promise than ship broken filtering. If Google adds a region filter, we'll wire it in.
**Why isn't there a search-by-keyword mode?**
Because Google's RPC doesn't expose one. You search by advertiser. For brand-keyword monitoring, give us the domain (e.g. `nike.com`) and the scraper returns every ad pointing at that domain β including those bought by competitors trying to bid on your name.
**Can I scrape political ads specifically?**
Not yet β political ads live in a separate Google library with its own endpoints. Open an issue on the Apify Store listing if you want this; we'll prioritize based on demand.
**How do I export to Google Sheets / a database?**
Three options:
1. **Console β Storage β Export** for one-off CSV downloads.
2. **Webhook** the dataset URL to a small Make / Zapier flow that appends to Sheets.
3. **Apify integration nodes** in Airbyte, n8n, or your warehouse loader.
**Some preview URLs are `null`. Why?**
Rich / video / animated creatives expose only a `content.js` URL β Google's preview is rendered by JavaScript. Static images give you a direct `preview_image_url`. Plan accordingly: if you need actual video frames, post-process the `content.js` URL with a headless browser.
**The number of returned ads is less than Google's reported total. Why?**
Google paginates and stops responding past some internal limit (we've observed ~1 000-ad caps per query). For very large advertisers (Nike, Coca-Cola), `maxPages = 25` is intentionally conservative. Raise it if you need full coverage.
### π¬ Support & feedback
Spotted a bug, missing field, or want a new feature? Open an issue on the [Apify Store listing](https://apify.com/DevilScrapes/google-ads-transparency) β we read every one.
Built by **Devil Scrapes** β Apify Actors with attitude. PPE, transparent pricing, no junk fields.
# Actor input Schema
## `searchDomains` (type: `array`):
List of advertiser landing-page domains to scrape. The scraper returns every ad pointing at each domain in the date window. Example: