What is MCP for Google Ads?
MCP is the Model Context Protocol — an open standard that lets an AI client call external tools and read external data in a structured way. Without it, an assistant has no route to your Google Ads account.
GoMarble MCP is an MCP server that connects supported AI clients to marketing data and workflows in GoMarble. For Google Ads that means reaching accounts, campaigns, ad groups, keywords, search terms, assets and change history, along with the performance data attached to each.
What makes Google different from paid social is that the account tells you more if you ask precisely. Search terms are exposed, keyword-level data exists, and the change log is readable — so most Google questions are really query questions. GoMarble MCP includes GAQL access, which means the assistant can compose the exact query a question needs rather than working from whatever a fixed report happens to include.
How to connect Google Ads to GoMarble MCP
Connecting Google Ads to GoMarble MCP takes a few steps. Once connected, you query your account from the AI tool you already use.
- Connect your Google Ads account to GoMarble.
- Connect GoMarble MCP to your preferred supported AI client.
- Authorize the access GoMarble requests.
- Choose the GoMarble workspace and accounts the client can use.
- Start asking questions or running supported workflows.
Connecting to Claude
The current steps, from the GoMarble MCP page:
- Open Settings, then Connectors.
- Choose Add custom connector.
- Enter https://apps.gomarble.ai/mcp-api/sse, then select Authorize.
- Choose the GoMarble workspace and accounts Claude can use.
Authorization uses GoMarble’s consent and workspace permissions, completed in the browser window your client opens.
Other supported clients
GoMarble MCP is a remote MCP server at https://apps.gomarble.ai/mcp-api/sse, currently supported in Claude, ChatGPT, Cursor, n8n, Hermes and OpenClaw, plus other clients that support compatible remote or local MCP servers. Setup differs slightly per client, so follow the canonical instructions rather than adapting the Claude steps.
View the complete GoMarble MCP setup guide →
What the free plan includes
- Up to $50,000 in cumulative ad spend
- One account per data source
- No edit actions — read and analysis only
What can you do with GoMarble MCP for Google Ads?
Grouped by the job rather than by the tool that does it.
Analyze
- Run GAQL queries against any reportable Google Ads resource
- Analyse account, campaign, ad group, keyword and ad performance
- Compare periods and identify trends
- Break down spend, ROAS, CPA, CPC, CTR, conversions and impression share
- Read account currency so figures are interpreted correctly
Diagnose
- Investigate why ROAS or CPA moved and localise it
- Mine search terms for spend with no conversion contribution
- Identify keywords carrying cost without return
- Read the change log to see what was edited before a shift
- Separate budget-limited from rank-limited delivery using impression share
Creative and asset intelligence
- Analyse ad and asset performance
- Inspect responsive search ad assets and their contribution
- Review Performance Max asset groups
- Compare messaging across ad variants
Research
- Discover keyword ideas for a topic, page or seed set
- Pull keyword metrics including volume and competition
- Research the competitive picture through the ads library
Recommend
- Surface negative keyword candidates from search-term data
- Recommend where budget is constrained versus wasted
- Suggest campaigns and ad groups worth restructuring
- Flag optimisation opportunities with supporting numbers
Act
- Propose campaign, ad group, ad and asset updates
- Propose bid modifier changes
- Propose a negative keyword list
- Propose a new campaign structure, experiment or PMax asset group
- Every write is a proposal that routes through your workspace's approval gates
GoMarble MCP capabilities for Google Ads
| Capability | What GoMarble MCP can do |
|---|---|
| GAQL querying | Compose and run queries against reportable Google Ads resources rather than relying on fixed reports |
| Performance analysis | Analyse spend, ROAS, CPA, CPC, CTR, conversions and impression share at any level |
| Search-term analysis | Mine search terms for waste and negative keyword candidates |
| Keyword research | Discover keyword ideas and pull volume and competition metrics |
| Performance diagnosis | Investigate the cause of a change and localise it to campaigns, ad groups or keywords |
| Asset analysis | Review ad assets and Performance Max asset groups |
| Change history | Read the account change log to correlate edits with performance shifts |
| Reporting | Generate account, campaign and keyword summaries on request |
| Actions | Propose updates, negative keyword lists, experiments and structures, subject to approval gates |
What data can GoMarble MCP access?
The data available for Google Ads, grouped by type.
Account data
Accounts and manager hierarchy · Account currency · Historical performance · Change log
Campaign structure
Campaigns · Ad groups · Ads · Budgets · Bid strategies · Bid modifiers
Keyword and query data
Keywords · Match types · Search terms · Keyword ideas · Search volume and competition
Performance metrics
Spend · Conversions and conversion value · ROAS and CPA · CPC and CTR · Impressions and impression share · Device and geography breakdowns
Assets and creative
Responsive search ad assets · Asset performance · Performance Max asset groups · Sitelinks and extensions
Experiments
Experiments · Experiment performance
Cross-platform
Meta Ads · TikTok Ads · LinkedIn Ads · Bing Ads · GA4 · Search Console · Shopify · Klaviyo
GoMarble MCP workflows for Google Ads
Jobs marketers actually want done, rather than features. Each one is a single natural-language request.
Mine search terms for waste
“Which search terms are spending without converting?”
What GoMarble MCP does: Queries the search term view for the period with spend above zero, joins cost and conversion data, and isolates terms carrying meaningful spend with no return.
What you get: A ranked list of negative keyword candidates sized in money, with the campaigns and ad groups they came through.
Diagnose a CPA increase
“What caused my Google Ads CPA to increase this week?”
What GoMarble MCP does: Compares the period against the prior one, separates cost-side from conversion-side movement, then drills by campaign, ad group, keyword, device and geography.
What you get: The mechanism behind the increase and where it originated, rather than the fact that it happened.
Check whether you are budget or rank limited
“Are my campaigns losing impression share to budget or to rank?”
What GoMarble MCP does: Queries impression share alongside lost impression share by budget and by rank at campaign level.
What you get: Which constraint is actually binding, which decides whether the answer is more money or better bids and quality.
Find keywords carrying cost without return
“Which keywords are spending the most with the worst conversion rates?”
What GoMarble MCP does: Queries keyword-level cost, clicks, conversions and conversion rate, normalising for volume so conclusions are not drawn from noise.
What you get: The keywords worth pausing, restructuring or re-matching, with the numbers behind each.
Correlate a change with an account edit
“Did anything change in the account before performance shifted?”
What GoMarble MCP does: Reads the change log for the window around the shift and compares it against the performance timeline.
What you get: The edits that preceded the change, which is the fastest route to a cause and the step most often skipped.
Research keywords for a new campaign
“Find keyword ideas for [topic] and tell me which are worth targeting.”
What GoMarble MCP does: Runs keyword discovery for the seed, pulls volume and competition metrics, and assesses intent against what you sell.
What you get: A candidate keyword set with volume and competition attached, grouped by intent.
Review Performance Max asset groups
“How are my Performance Max asset groups performing?”
What GoMarble MCP does: Reads asset groups and their performance, including asset-level contribution where available.
What you get: Which asset groups and assets are carrying the campaign, on a surface that exposes limited controls by design.
Generate a client-ready summary
“Summarise this account's performance this month against last.”
What GoMarble MCP does: Queries account and campaign performance for both periods and surfaces the notable movements with causes.
What you get: A summary with the changes that matter and why they happened.
Example GoMarble MCP prompts for Google Ads
Things real marketers ask, grouped by intent. Paste any of them once connected.
Performance analysis
- “How did this account perform last month versus the month before?”
- “Show me spend and conversions by campaign this month.”
- “What is my CPA by ad group?”
- “How is impression share trending on my top campaigns?”
Diagnosis
- “Why did my CPA increase this week?”
- “Which campaigns are responsible for the spend increase?”
- “Am I losing impression share to budget or to rank?”
- “Did anything change in the account before performance dropped?”
- “Which devices or locations are dragging performance down?”
Search terms and keywords
- “Which search terms are spending without converting?”
- “Find negative keyword candidates from the last 30 days.”
- “Which keywords have the highest cost and lowest conversion rate?”
- “Are any of my keywords competing with each other?”
- “Find keyword ideas for [topic].”
Budget and spend
- “Which campaigns are budget-constrained?”
- “Where does my budget allocation look inefficient?”
- “How much am I spending on branded versus non-branded terms?”
Assets and creative
- “Which responsive search ad assets are performing best?”
- “How are my Performance Max asset groups doing?”
- “Compare messaging across my ad variants.”
Reporting
- “Summarise this account for the month.”
- “Build me a keyword-level breakdown I can send to a client.”
Actions
- “Propose a negative keyword list from the wasteful search terms.”
- “Propose bid modifier changes for the devices that are underperforming.”
- “Propose pausing the keywords with spend and no conversions.”
Why use GoMarble MCP for Google Ads?
An API gives AI access to data. GoMarble MCP gives it the marketing context and workflows needed to work with that data.
GAQL, composed for the question
The Google Ads API is powerful and verbose. GoMarble MCP lets the assistant compose the query a question actually needs, which is the difference between a fixed report and an answer.
Search terms are first-class
Most recoverable Google waste hides in search terms. They are directly queryable here, alongside the cost and conversion data that makes them actionable.
Change history joins the analysis
Correlating a performance shift with what someone edited is usually the fastest diagnosis available, and it needs both datasets in one place.
Writes are governed
Changes are proposed and routed through your workspace's approval gates rather than issued directly with whatever the token permits.
Search is not the whole funnel
The same connection reaches Meta, TikTok, GA4, Shopify and Klaviyo, so a Google result can be read against the rest of the acquisition picture.
GoMarble MCP for cross-platform marketing analysis
Marketers rarely decide anything from one platform. GoMarble MCP can give AI a view of the wider marketing system rather than forcing it to reason from one disconnected source at a time.
| Combination | What becomes answerable |
|---|---|
| Google Ads + Meta Ads | Compare search and paid social on a consistent basis, and catch the interaction where cutting social prospecting raises brand-search CPA. |
| Google Ads + Search Console | See paid and organic query performance together, which is how you find terms you are paying for and already rank well on. |
| Google Ads + GA4 | Join click data to on-site behaviour and conversions to separate an ad problem from a landing-page problem. |
| Google Ads + Shopify | Put spend against store revenue and product-level performance rather than platform-reported conversions alone. |
| Google + Bing | Compare search platforms directly when budget spans both. |
How does GoMarble MCP work?
| Step | What happens |
|---|---|
| 1 | You ask a question in natural language |
| 2 | The AI client works out what information it needs |
| 3 | GoMarble MCP retrieves the relevant marketing data and tools |
| 4 | GoMarble supplies the marketing context around it |
| 5 | The AI analyses the result |
| 6 | You get an answer, a recommendation, a report — or a proposed action |
Write access is plan-dependent. On the read-only plan you can connect a client and analyse supported accounts without making changes. Where write access is enabled, changes are proposed rather than applied directly — every write tool in the server is a propose call, and consequential actions route through the permissions and approval gates configured for your workspace.
Who is GoMarble MCP for?
PPC managers
Query the account precisely without writing GAQL by hand or waiting on a report build.
Search specialists
Mine search terms and keyword data conversationally, which is where most recoverable waste lives.
Agencies running many search accounts
Run the same investigation across accounts and produce client-ready summaries from it.
In-house acquisition teams
Diagnose a CPA move and check whether budget or rank is the constraint in one sitting.
Ecommerce teams running Shopping and PMax
Read product and asset group performance against store revenue.
GoMarble MCP vs traditional Google Ads workflows
| Traditional workflow | With GoMarble MCP |
|---|---|
| Build a report or write GAQL by hand | Ask the question and let the assistant compose the query |
| Download the search terms report and filter it | Ask which terms are wasting spend |
| Check the change history in a separate tab | Ask what changed before performance moved |
| Export to compare periods | Compare periods in the conversation |
| Repeat the same weekly keyword review | Re-run the same prompt |
| Build a negative keyword list manually | Propose the list from the data, then approve it |
Frequently asked questions
What is MCP for Google Ads?
What is GoMarble MCP?
Can GoMarble MCP run GAQL queries?
Can GoMarble MCP analyse search terms?
Can GoMarble MCP explain why Google Ads performance changed?
Can GoMarble MCP identify wasted Google Ads spend?
Can GoMarble MCP make changes to my Google Ads account?
Is GoMarble MCP read-only?
What are the free plan limits?
Which AI assistants work with GoMarble MCP?
How do I connect Google Ads to GoMarble MCP?
Does it support keyword research?
Can it analyse Performance Max campaigns?
How is this different from using the Google Ads API directly?
Why GoMarble MCP?
The work you stop doing:
- Stop writing GAQL by hand to answer a one-off question
- Stop downloading the search terms report to filter it in a spreadsheet
- Stop checking the change log in a separate tab
- Stop rebuilding the same weekly keyword review
- Stop guessing whether budget or rank is the real constraint
Connect your marketing stack once. Give AI the context it needs. Ask in plain English, and let it analyse, explain, recommend and — where enabled — propose the change.
Your Google Ads data is already there. Now give AI access to it.
Connect GoMarble MCP to your marketing stack and start analysing Google Ads from the AI tool you already use.