---
title: "Google Ads MCP: Transparency Center tools and limits"
description: "How Proxy's Google Ads MCP reads the Ads Transparency Center: advertiser lookup, ad lists, creative details, exact parameters, paging and credit cost."
---

> For AI agents: the full docs index is at [https://useproxy.dev/docs/llms.txt](https://useproxy.dev/docs/llms.txt). Append `.md` to any docs URL for Markdown.

Most servers called a "Google Ads MCP" manage your own Google Ads account. Proxy does a different job: it reads the public Google Ads Transparency Center, so an assistant can show what any advertiser runs on Search, YouTube and Display. It is read-only and never connects to a Google Ads account.

This page documents exactly how the three Google tools behave. For the short parameter tables, see the [Google tool reference](https://useproxy.dev/docs/tools/google-ads).

:::note
Google's library is searched by advertiser or domain. There is no keyword search inside ad copy, and no spend data.
:::

## Connect

One endpoint serves every client and every platform:

```text
https://mcp.useproxy.dev/mcp?ref=docs
```

Sign-in is OAuth, with no API keys. Steps per client are in [Connect your client](https://useproxy.dev/docs/clients), with deeper notes for [Claude](https://useproxy.dev/docs/ads-mcp-claude), [ChatGPT](https://useproxy.dev/docs/ads-mcp-chatgpt) and [Claude Code](https://useproxy.dev/docs/ads-mcp-claude-code).

## The three-step chain

Google research always runs in the same order:

1. `get_google_platform_id` turns a brand name into advertiser IDs and domains.
2. `get_google_ads` lists the ads for one advertiser ID or one domain.
3. `get_google_ad_details` opens an ad to read its copy, regions and impressions.

The list step returns no ad copy. Headlines and descriptions only exist at step 3.

## get_google_platform_id

| Parameter | Type | Behaviour |
| --- | --- | --- |
| `brand_name` | string or list | Bare brand name or bare domain. A list is looked up in parallel |

- **Matching is literal.** Pass `Purina` or `purina.com.mx`, not both together, and no country or city.
- **Built-in fallbacks.** If the exact query finds nothing, the server retries with the bare domain, the bare name, or the name without a trailing country. The response then says which query produced the results. These retries cost nothing extra.
- **Two result types.** `advertisers` (name, `advertiser_id`, region) and `websites` (domain). They can point to different accounts for the same brand, so the assistant is told to query both.
- **Results span all countries.** Pick the advertiser whose region matches your market.
- **Zero matches** returns a hint to retry with the bare name or domain. Unlike Meta, there is no keyword fallback.

## get_google_ads

| Parameter | Default | Accepted values and behaviour |
| --- | --- | --- |
| `domain` | none | For example `nike.com`. Company-wide search |
| `advertiser_id` | none | From the lookup. Targets one regional entity |
| `limit` | 100 | Maximum 1000. Fetched in pages, up to 10 pages per call |
| `topic` | none | Topic filter. `political` requires `region` |
| `region` | none | Region code such as `US` |
| `cursor` | none | Reads the next window of a result already fetched |

One of `domain` or `advertiser_id` is required. Calling with neither, or with `topic` `political` and no `region`, returns an error.

Each ad row carries `ad_url`, `advertiser_id`, `advertiser_name`, `format`, `domain`, `first_shown` and `last_shown`. `format` is text, image or video.

Re-running the same advertiser later reports what changed since your last fetch: how many ads are new and how many are no longer returned.

## get_google_ad_details

| Parameter | Type | Behaviour |
| --- | --- | --- |
| `ad_url` | string or list | Transparency Center ad URL from `get_google_ads` |

A single URL returns:

- `advertiser_id`, `creative_id`, `format`, `first_shown`, `last_shown`
- `overall_impressions`, which can be empty
- `regions`, the region codes the ad is shown in
- `variations`, each with `headline`, `description`, `destinationUrl`, `allText` and, for image ads, `imageUrl`

A list of URLs is fetched in parallel and returns a summary per URL: `format`, `variation_count` and `image_urls`. To read the copy of a specific ad, ask for that ad on its own.

Image URLs can be passed to [analyze_ad_image](https://useproxy.dev/docs/tools/analysis#analyze_ad_image).

## What it never contains

- Spend, clicks, CPC or conversions. Google does not publish them.
- The keywords an ad bids on.
- Impressions at list level. Ranges appear only in details, and can be empty there too.
- Anything from your own Google Ads account.

## Paging and cursors

- `get_google_ads` pages upstream until it reaches `limit` or 10 pages.
- A large list is returned in windows of about 40,000 characters. The rest is stored for 24 hours and read with a cursor of the form `rs:<id>:<offset>`, at no credit cost.
- After 24 hours the cursor returns an error asking to re-run the call.

## Credit cost

| Call | Credits |
| --- | --- |
| `get_google_platform_id` | 1 per brand, including any fallback retries |
| `get_google_ads` | 1 per upstream page fetched, up to 10 |
| `get_google_ad_details` | 1 per ad URL. Duplicate URLs in a list are fetched once |
| Reading an `rs:` cursor | 0 |

`limit` is the cost lever. Ask for 20 to 30 ads for a quick look and keep 100 for a full audit.

Free is 20 credits a day with no card, Pro is $50 a month for 100 a day, and Max is $100 a month for 250 a day. There is no overage billing. See [Credits and plans](https://useproxy.dev/docs/credits).

## Failure modes

| Situation | What the tool returns |
| --- | --- |
| Name with a country or city attached | Results for the fallback query, with a note naming it |
| One brand in a list fails | That brand is marked "lookup failed, not confirmed absent" |
| No ads for the identifier | Zero ads. Try the other identifier type before concluding |
| Ad removed since it was listed | "Google returned no details for this ad_url" |
| One URL in a list fails | That URL carries an error; the others still return |
| Credits run out mid-list | The first URLs are fetched; the rest are marked as not fetched |
| Rate limit, timeout or outage | A short retry hint. Raw upstream errors are never passed through |

## Worked prompts

```text
Find [brand]'s Google advertiser accounts and domains.
```

`get_google_platform_id`. 1 credit.

```text
Pull 30 Google ads for [domain] and group them by format.
```

`get_google_ads` with `domain` and `limit` 30. Usually 1 credit.

```text
Show every headline and description variation in [brand]'s newest text ad.
```

`get_google_platform_id`, `get_google_ads`, then `get_google_ad_details` with one `ad_url`. 3 credits.

```text
Which regions is [brand] showing its five longest-running Google ads in?
```

`get_google_ads`, then five single `get_google_ad_details` calls for `regions`. 6 credits plus the lookup.

```text
Find three competitors of [domain] and compare their Google ad formats.
```

`search_competitors`, then one `get_google_ads` per domain. About 4 credits at 30 ads each.

## Guides

- [Google Ads Transparency Center guide](https://useproxy.dev/blog/google-ads-transparency-center)
- [How to check competitors' Google ads](https://useproxy.dev/blog/how-to-check-competitors-google-ads)
- [Google Ads Transparency Center API options](https://useproxy.dev/blog/google-ads-transparency-center-api)
