Quattr leads AEO, SEO, and content rankings on G2 Spring 2026. View our G2 badges →
Request demo
Request demo

Updated Aug 8, 2026

4 best Claude MCPs & ChatGPT plugins for Google Ads (2026)

Paid search data in an AI tool runs from raw GAQL to governed cross-channel analysis. GAQL is Google Ads Query Language, the API's SQL-like reporting language. These MCPs put your Google Ads accounts, spend and performance behind your assistant. They also put paid and organic on the same evidence standard.

Ranked on

  1. First-party grounding Does it answer from accounts you own: Search Console, analytics, ads, logs? Or from a third-party index and modeled estimates?
  2. Governed analyses & statistics Does the server run defined analyses with significance gates? Those gates stop a thin sample reading as a confident answer. Or does it hand your assistant raw rows to assemble and judge itself?
  3. Provenance on every answer Does each answer carry its scope and a freshness date? Can you click back to the source?
  4. Write scope and spend guardrails Can the server change campaigns or budgets? Settle that before you connect credentials that can spend money. Ask whether write tools can be denied or scoped.
  5. Paid-organic join on named goals Named goals are the conversions you defined in your own analytics. Can ads data sit beside Search Console and those goals in one answer? Or is joining the channels left to you?

Our entry is in this list. We publish it, and rank 1 is ours, so read the criteria first and the competitor entries on their own terms. Each is described from its public documentation, with what it does better than us stated where it does.

How this was checked

Verification window
Aug 2 to Aug 8, 2026
Read for every entry
  • each vendor's own product documentation
  • release notes and changelogs
  • published pricing and metering pages
  • documented client-support matrices

If a vendor's own docs don't mention a feature, we mark it "not documented." That's a gap in what's published, not a verdict that the feature is missing. Where their docs contradicted a claim we carried, we changed the claim.

What is in, and what is not

Qualified for this list

  • It puts Google Ads data behind an MCP client. That means your own accounts through the Ads API, or Google Ads-sourced keyword metrics.
  • Documentation had to be fetchable and dated, covering auth, data scope and tool behaviour. A server that inherits that surface from the Google Ads API itself counts.

Looked at and left out

  • Hosted third-party wrappers of the Google Ads API, They resell the same API surface; we ranked the first-party server and the self-hosted class instead.
  • GA4-only analytics MCP servers, They carry sessions and conversions but not ad spend, campaign structure or auction coverage.
  • Third-party paid-search estimate panels, They model competitors' ad spend and copy from panels rather than reporting your account's billed spend.

The ranking

  1. 1 Quattr MCP + plugin Ours, bias disclosed

    Best for: paid & organic in one governed view

    58 governed analyses over your first-party data. Governed means pre-built analyses with locked definitions. The same question always computes the same way. Google Ads sits beside GSC, GA4/Adobe named goals and AI-agent logs. Quattr's own SERP tracking observes ads in the results too. One control plane for paid and organic.

    • Your own sources: GSC (incl. BigQuery), GA4/Adobe named goals, Google Ads, Cloudflare and other AI-agent logs. Concierge-onboarded.
    • Quattr's own classic SERP tracking observes ads in the results, beside your account data
    • Significance tests return a verdict on whether a change is real. Refuses causal claims from a trend line.
    • Interactive cards carry scope, freshness, and deep-links. An ⚠ Observational chip marks numbers as observed, not causal. If a source is missing, it refuses and gives the reason.

    Consider

    • The analyses are fixed. You compose questions against them. There is no arbitrary GAQL or SQL over your ads data.
    • Included with a Quattr subscription. There is no standalone tier for the MCP alone.
    • Read-only across connected accounts. It reports and diagnoses. It cannot change bids, budgets or campaign status.

    Works in: Claude · Claude Code · ChatGPT · Cursor · VS Code · Goose Included with a Quattr subscription. No per-call metering.

    Verified against Quattr MCP docs, Aug 2026

    Request a demo

  2. 2 Google Ads API MCP server

    Best for: raw GAQL, first-party & free

    Google's official open-source Ads API MCP server executes GAQL against your own accounts. The current release is read-only. The API's discovery document, metrics and segments are exposed as resources. That document is a machine-readable map of the API's fields. The agent writes valid queries instead of guessing.

    • First-party, read-only GAQL access from the Ads API team itself
    • Free and Apache-2.0. Run it locally via pipx, or self-deploy to Cloud Run.

    Consider

    • Read-only in the current release. Google's docs state it cannot modify bids, pause campaigns or create assets.
    • You host it and you authenticate it. That means a Cloud project, OAuth credentials and a developer token. A token at test-access level reaches only test accounts.
    • Scope is the Ads API alone. No Search Console, analytics or named-goal joins. It returns GAQL rows.

    Works in: Any MCP host (Gemini CLI documented) Requires OAuth credentials and a Google Ads developer token. You host it.

    Verified against Google Ads API developer docs, Ads Developer Blog and the googleads GitHub README, Aug 2026

    Their docs ↗

  3. 3 DataForSEO MCP

    Best for: ads-side research on any market

    DataForSEO's official open-source MCP includes their Keywords Data module. That module returns Google Ads to based search volume and keyword metrics, per their docs. SERP, Labs and eight other raw data modules sit beside it.

    • Google Ads to based keyword data without connecting an ads account
    • Open source, with pay-as-you-go billing from a $1 non-expiring trial credit

    Consider

    • Keyword metrics come from DataForSEO's own Google Ads API access. Your ad account is not connected, and it does not need to be. So there is no spend, campaign or conversion data.
    • Metered per API call against prepaid credits. The pricing page states a $50 minimum payment amount.
    • Google Ads keyword data refreshes mid-month, per their docs. Live endpoints cap at 12 requests per minute.

    Works in: Claude Desktop · Claude Code · ChatGPT · Cursor · Gemini CLI · n8n Pay-as-you-go API credits, $50 minimum top-up. No subscription required.

    Verified against DataForSEO MCP page, pricing page, Keywords Data Google Ads docs and GitHub README, Aug 2026

    Their docs ↗ Quattr MCP vs DataForSEO MCP →

  4. 4 BigQuery MCP (Google)

    Best for: warehouse-scale ads analysis

    Google ships two official BigQuery MCP options. One is a managed remote endpoint. The other is the open-source MCP Toolbox. Both run SQL and schema exploration over your own BigQuery datasets. That is wherever your ad-platform exports land.

    • A fully managed remote endpoint with no server to run, plus an open-source local path
    • IAM-scoped OAuth with per-action roles and audit logging documented

    Consider

    • The execute_sql tool is not read-only. Google documents restricting it with a deny policy. For read-only work there is execute_sql_readonly.
    • It carries no ads semantics of its own. Your Google Ads data has to land in BigQuery first, through an export you maintain.
    • Results cap at 3,000 rows. Queries are auto-cancelled after three minutes, per Google's documented tool constraints.

    Works in: Gemini CLI · ChatGPT · Claude · Cursor · VS Code Free tooling. BigQuery query costs apply to usage.

    Verified against Google Cloud BigQuery MCP documentation, Aug 2026

    Their docs ↗

Side by side

Every entry against the same five criteria. Not documented means the vendor's own material does not address it, a gap in what is published, not a judgement about what the product can do.

Product First-party groundingGoverned analyses & statisticsProvenance on every answerWrite scope and spend guardrailsPaid-organic join on named goals
Quattr MCP + plugin paid & organic in one governed view Your connected accounts, Google Ads, Search Console, GA4/Adobe; no third-party index. Governed and fixed, significance-gated; refuses rather than returning an underpowered number. Cards carry scope, freshness and deep-links, refusals name the missing source. Read-only, reports and diagnoses; cannot change bids, budgets or campaign status. One analysis, ads beside Search Console and your own named goals.
Google Ads API MCP server raw GAQL, first-party & free Your own Ads accounts via GAQL, OAuth and developer token required. GAQL execution only, no analyses or significance testing documented. Not documented Strictly read-only, cannot modify bids, pause campaigns or create assets. Ads API scope only, no Search Console or analytics goal joins.
DataForSEO MCP ads-side research on any market DataForSEO's own Google Ads API access, not your ad account. Data endpoints return metrics, no analyses or significance testing documented. Not documented Research endpoints only, no ads account connected, so no spend changes. No account data, keyword metrics only, no spend, conversions or goals.
BigQuery MCP (Google) warehouse-scale ads analysis Your BigQuery datasets, ads data must be exported there first. SQL execution and schema tools, no significance testing documented. Not documented execute_sql permits mutations, restrict via deny policy; execute_sql_readonly stays read-only. Only what you export, joins are yours to write in SQL.

Self-hosted servers, and why they are not ranked

Community servers over the Google Ads API came first. cohnen/mcp-google-ads was published in March 2025, more than six months before Google shipped its own. It is still the best-known of them. It had gone ten months without a commit when we checked in August 2026. Others have arrived since, including multi-platform servers that put Google Ads in the same tool list as Meta, Amazon and TikTok. GitHub returns hundreds of matches for the category, and no two of them make the same promises.

The case for self-hosting narrowed once Google published a first-party server. The main draw was raw GAQL access against your own token, with nobody in between. That is now available, and read-only. What is left is extensibility. You can fork a repository, add tools to it, and run the thing inside your own network. If your team is going to write code against the Ads API anyway, someone's MIT-licensed server is a reasonable head start.

The costs do not disappear when the software is free. Every server on the Ads API needs your own developer token and OAuth credentials. A token with test access reaches only test accounts, so basic-access approval is a real step before any of this returns live numbers. You host it. You rotate the credentials. You own the upgrade when the API version moves. There is no support channel and no uptime commitment behind a repository.

This class is not scored in the table because the criteria ask what a publisher documents, and here there is no publisher. Write scope makes that concrete. It is also the thing to settle before a community server ever sees credentials that can spend money. It splits three ways. Google's own server documents itself read-only, which is part of why it ranks second. Some community servers do write, and say so. amekala/ads-mcp covers the full campaign lifecycle across six ad platforms. It documents its own safeguard too: new campaigns are always created paused. And some document neither way. Treat that third case as unsafe, because a server built on the Ads API can request mutate access whether or not its README mentions it. Read the tool list before you connect. Verbs like create, update and pause tell you what it can do to live spend. No read-only claim is not the same thing as a read-only server.

Test one of these in an afternoon

  1. Connect one server to a read-only Google Ads login. Then ask for last month's cost and conversions by campaign.
  2. Ask which of those conversions maps to a named goal in your analytics. Then ask how organic did on the same queries.
  3. A trustworthy answer names the account and the date range. It states how fresh the data is. It links back to the report or row it came from.
  4. Then ask for something it cannot have: a competitor's actual ad spend. It should refuse and name the missing source. An estimate is the wrong answer.

Anticipated questions

Is this list biased?
Yes. It's Quattr's site and Quattr ranks first. That's why the bias sits at the top of the page. The criteria are stated. Every third-party claim is verified against public documentation, with a date. Judge the criteria, not our word.
Is Google's official Ads MCP enough?
For read-only GAQL access to your own accounts, yes. It is first-party and free. You run it yourself, with OAuth credentials and a developer token. What comes back is raw rows. The analysis is a different job. Quattr joins Google Ads to organic and your named goals, through governed and significance-gated analyses. The BigQuery routes serve warehouse-scale work.
Where do community and self-hosted servers fit?
We don't rank them any more. Community servers over the Google Ads API predate Google's official option, and they are still widely used. That is why they get the prose section above instead of a scored row. We don't evaluate them one by one. Check maintenance, auth model and data scope yourself before you trust one with ads credentials.

Written by

Mahi Kothari Senior Content Strategist, Quattr

Works at the intersection of content strategy, technical SEO and AI visibility. Covers answer-engine optimization, generative engine optimization and what it takes for a brand to be cited in AI answers rather than only ranked.

Change history

  1. Aug 19, 2026 Fact-audit pass. This list's Aug 17 pass was prose-only, so every claim here was re-verified against vendor documentation for the first time since Aug 8. Corrected one claim that had gone stale: amekala/ads-mcp was described as covering four ad platforms, but its README now documents six (Google, Meta, LinkedIn, TikTok, Amazon and ChatGPT Ads). That repository was updated after this page's as-of date, so the original count was right when written and is not any more. Everything else re-verified and unchanged. Google's Ads API MCP server is still strictly read-only in the current release, in its own words. The BigQuery MCP row cap of 3,000, the three-minute query timeout, execute_sql being the one tool that is not read-only, and the execute_sql_readonly alternative all still hold. DataForSEO's live Google Ads endpoints still cap at 12 requests per minute, and its Google Ads keyword data still refreshes mid-month. A test-access developer token still reaches only test accounts. The cohnen/mcp-google-ads dates were checked against the repository itself: first published March 2025, last commit October 2025, which is the ten-month gap described.
  2. Aug 17, 2026 Moved community and self-hosted servers out of the scored table: a category has no single documentation set, so per-criterion scoring could not be true or false. The class is now covered in prose, and the ranked count drops to four. Rewrote the prose for readability against the house voice guide. No ranking, criterion or verified claim changed in that pass.
  3. Aug 8, 2026 Re-verified every written entry against vendor documentation. Added documented considerations to each one including our own, ranking criteria specific to this list with a definition apiece, the verification window and sources, inclusion criteria, and a side-by-side table.
  4. Aug 2, 2026 First publication of the twelve lists.

Request a demo Take a test drive All 12 lists

Third-party details as of Aug 8, 2026; marks belong to their owners. Spot an error? Tell us and we'll fix it: info@quattr.com