Claude can read your Meta Ads and Google Ads, but there's no single official route. Meta has its own connector, added by URL and still in beta. Google's official servers for Google Ads and Google Analytics only read, and run on your own computer. For questions that span channels, an aggregator such as Supermetrics, Windsor.ai or Funnel is simpler: one connector, every platform, one set of definitions. Go direct when a question names one platform.
This is the performance analysis kit in our series on Claude connectors for each marketing role. If you want Claude to change campaigns rather than report on them, see the media buying kit.
Is there an official Claude connector for Google Ads and Meta Ads?
Each platform has answered differently, and only TikTok's sits in Claude's directory. Here's where each one stood when we checked on 9 November 2026.
What's official for each ad platform
| Platform | Official route | How you add it | Can it change things? |
|---|---|---|---|
 Meta Ads | Meta's Ads MCP server, in open beta since 29 April 2026. 29 tools across reporting, campaigns, catalogues and pixel checks. | Paste mcp.facebook.com/ads into Claude as a custom connector, then sign in with Facebook. | Yes: read and write |
 Google Ads | Google's open-source Google Ads MCP server. | Install it on your computer or your own cloud, with a Google Cloud project and a developer token. | No: read only |
 Google Analytics 4 | Google's open-source Analytics MCP server, labelled experimental. | Install it on your computer, signed in with a read-only Google Analytics scope. | No: read only by default |
 TikTok Ads | TikTok for Business, made by TikTok. Campaigns, reporting, audiences and creative. | Add it from Claude's connector directory. | Yes: read and write |
| LinkedIn and Microsoft Ads | No ads connector from either. | Through an aggregator. | Depends on the aggregator |
Checked against Meta for Developers, Google's developer docs and Claude's connector directory on 9 November 2026. Installed on your computer, Google's servers work in Claude Desktop and Claude Code, not in Claude on your phone.
What do aggregators do differently?
An aggregator connects to your ad and analytics platforms once, then gives Claude a single connector that can query all of them in the same language. Five build their own Claude connectors. You need one.
1. Supermetrics: reports, dashboards and monitors
Supermetrics queries more than 200 platforms, including Google Ads, Meta, LinkedIn, TikTok, GA4, Search Console and Shopify, and can manage campaigns and social posts. It shares Claude-built dashboards as live links and runs monitors on your data. It has a 14-day trial, then needs a subscription.
What you can do with it
- Build a cross-channel weekly report and share it as a live dashboard.
- Set a monitor for spend or cost-per-acquisition spikes.
- Pull Instagram insights next to paid social in the same table.
Try: “Compare spend, cost per acquisition and return on ad spend across Meta, Google Ads and TikTok for the last 30 days, and turn it into a dashboard I can share.”
2. Windsor.ai: the most sources, and it can act
Windsor.ai covers more than 350 sources and can change campaigns, posts and listings, including Google Ads. Its free plan covers one source; more start at $19 a month. Its own example is the question we'd ask first: Meta and Google Ads spend against GA4 sessions and Shopify or HubSpot revenue, in one go.
What you can do with it
- Join ad spend to GA4 sessions and store revenue in one table.
- Compare one ad account with several client accounts.
- Act on what the report shows, with each change confirmed.
Try: “Compare Meta and Google Ads spend with GA4 sessions and Shopify revenue by week for the last quarter.”
3. Funnel: analysis with forecasts
Funnel is built for analysis across more than 600 platforms, with your own business context and definitions on top. If you use its measurement features, the connector also runs saturation curves, forecasts and what-if scenarios. It has separate US and EU servers, which matters if your data has to stay in Europe.
What you can do with it
- See where each channel starts to saturate.
- Forecast next quarter's results from the current mix.
- Ask which campaigns generated the most HubSpot pipeline.
Try: “If we move 15% of Meta budget to Google Search next month, what happens to revenue?”
4. Polar Analytics: for e-commerce brands
Polar connects Shopify, Meta, Google, Klaviyo and more than 45 other sources into one layer with validated metrics, your own business rules and cross-platform attribution. The point is consistency: the same definitions and exclusions apply to every question, so you stop getting a different number each time. It handles multi-store set-ups.
What you can do with it
- Get blended ROAS, customer acquisition cost and margin by store.
- Rank SKUs by profit, not revenue.
- Compare attribution across platforms with the same rules.
Try: “What were blended ROAS, CAC and contribution margin last month for each store, and which SKUs made the most profit?”
5. Triple Whale: what changed, and why
Triple Whale is also built for e-commerce: summary KPIs, pixel attribution and an explanation of what changed, scoped to your store, timezone and currency. Claude can also run SQL against its tables.
What you can do with it
- Explain why blended return on ad spend dropped yesterday.
- Compare first-party pixel attribution with what each platform claims.
- Pull store KPIs for the weekly report.
Try: “Why did our blended ROAS drop last week? Break it down by channel and compare Triple Whale's attribution with each platform's own numbers.”
Should you use one connector per platform or an aggregator?
Use Intersect's one-question test. Count the platforms in the question you ask most often. One platform: go direct. Two or more: aggregate. If the answer leads to changing spend, use whichever route you can audit, and keep every change on Needs approval.
What about site and product analytics?
Ad platforms stop at the click. What happens on your site comes from a different set of connectors.
6. Google Analytics: Google's own server
Google's open-source Analytics server runs on your own computer and reads your GA4 properties: account and property details and reports. It's labelled experimental and is read only.
What you can do with it
- Report sessions and conversions by channel and landing page.
- Compare this month with the same month last year.
- List the accounts and properties you can report on.
Try: “Which landing pages had the most sessions from paid social last month, and what was their conversion rate?”
7. Contentsquare: what lost the conversion
Contentsquare captures every visitor action without tagging. Its connector computes funnels, journeys and page comparisons, and ranks errors and page groups by the conversions they cost you.
What you can do with it
- Find the errors costing the most conversions.
- Map the journey from a campaign landing page to checkout.
- Compare how two pages perform for the same goal.
Try: “Which errors on our site cost us the most conversions last month, and on which pages?”
8. Fullstory: the sessions behind the number
Fullstory gives Claude both halves of an investigation: analytics to size a problem, and session replay to see why it happened. Every answer links back to the sessions in Fullstory, and it catches silent failures that never threw an error, so “conversion dropped 12%” becomes “the coupon field fails for logged-out users”.
What you can do with it
- Find where checkout is leaking, and watch the sessions behind it.
- Explain why a metric moved after a release.
- Set a scheduled check for frustration on key flows.
Try: “Conversion from basket to checkout dropped this week. Find where, for whom, and show me sessions that explain it.”
9. Mixpanel: product analytics
If your conversion happens inside a product (sign-up, activation, upgrade), Mixpanel's connector runs segmentation, funnel and retention analyses, explores your event list, manages its definitions, analyses session replays and finds data quality problems.
What you can do with it
- Compare trial-to-paid conversion by acquisition channel.
- See retention for users who came from a campaign.
- Find events that stopped firing after a release.
Try: “Compare trial-to-paid conversion and 30-day retention for users from paid search, paid social and organic.”
10. Adobe Customer Journey Analytics: for Adobe shops
For companies on Adobe, the Customer Journey Analytics connector queries metrics, dimensions and segments across channels, runs reports and can create segments and calculated metrics, so the analysis follows your own methodology rather than a generic one.
What you can do with it
- Run a cross-channel journey report without opening Workspace.
- Create a segment and a calculated metric from a description.
- Find which components a project already uses.
Try: “Report conversion rate by marketing channel for visitors who used the store locator, and save that audience as a segment.”
Also connected: Microsoft Clarity gives Claude your Clarity analytics and session recordings, and Amplitude and PostHog cover product analytics too.
Why do two connectors disagree about the same sale?
Because each platform counts its own contribution, by its own rules. By default, Meta credits a sale that happens up to seven days after someone clicks an ad, or one day after they only saw it. Google Ads counts conversions up to 30 days after a click. GA4 doesn't count ad views at all, and shares credit across every channel that touched the sale. Ask Claude to add up what each platform reports and you'll count the same order more than once.
One sale, three claims: an example
Shopify orders1Meta reports1Google Ads reports1A shopper sees an Instagram ad on Monday, searches your brand on Wednesday, clicks a Google ad and buys. Both platforms are right by their own rules. Only the store knows there was one order.
The fix is a sentence in your instructions, not a new tool. Tell Claude which system is the truth for revenue (usually your store or CRM), to report platform numbers as each platform's claim, and never to add them up. An aggregator makes the comparison easier. It doesn't make the platforms agree.
Which should you choose?
- Mostly Meta: Meta's own connector. It's free, it can act, and the connection is between you and Meta.
- Mostly Google, with a technical person: Google's servers for reporting, read only. Add an aggregator when the questions start crossing channels.
- Three or more channels, or an agency: an aggregator. One connector, one set of definitions, and every client's accounts in one place.
- You want Claude to change budgets: start with pausing and reporting, keep every change on Needs approval, and check the platform afterwards. The media buying kit covers which connectors can.
Try: “Compare spend, cost per acquisition and return on ad spend across Meta, Google Ads and TikTok for the last 30 days. Use Shopify for revenue and show each platform's own reported conversions separately.”
Where does this break?
- Meta's connector is still a beta. Some agency write-ups report it saying ads tools aren't enabled yet for some ad accounts. Test it on your own account before you plan around it.
- Unofficial tools can cost you the account. Agencies reported Meta restricting ad accounts after connecting AI tools through unofficial routes. Meta's own connector is the route it approves.
- Write access is real money. A misread instruction that doubles a budget is spent before anyone notices. Keep changes on Needs approval.
- Local servers stay on one computer. Google's servers run where you install them, so they won't be there when you open Claude on your phone.
- Big tables fill the conversation. Ask for summaries over a set date range rather than every row from every campaign.
Want your channels in one place?
We'll choose the route for your ad accounts, connect them, write the reporting rules that stop Claude double-counting, and decide with you what it may change without asking.
FAQs
Can Claude analyse my Meta and Google Ads?
Yes. Connect Meta's own connector and Google's Google Ads server, or one aggregator that covers both. For a question across both platforms, the aggregator is less work.
Is there an official Meta Ads connector for Claude?
Yes. Meta launched one on 29 April 2026 and it's in open beta. It isn't in Claude's directory: you add mcp.facebook.com/ads as a custom connector and sign in with Facebook.
Can Claude change my campaigns?
Through Meta's connector, TikTok's, Supermetrics or Windsor.ai, yes. Google's official Google Ads server only reads. Keep any change on Needs approval.
What's the best way to report across channels in Claude?
An aggregator such as Supermetrics, Windsor.ai or Funnel, plus one instruction: treat your store or CRM as the truth for revenue, and show each platform's own conversions separately.
Is there a LinkedIn Ads connector?
Not from LinkedIn. Supermetrics, Windsor.ai and Funnel all include LinkedIn Ads.
Related reading
- The 10 Claude Connectors Every Team Should Set Up First
- Connectors: Give the Claude and ChatGPT Brain Your Business
- Your Data Isn't Ready for AI. Here's the Minimum It Needs.
Sources
- Meta for Developers: Ads MCP server overview, accessed 9 November 2026
- MCP Directory: Inside Meta's official AI connectors, accessed 9 November 2026
- Google Ads API: MCP server developer guide, accessed 9 November 2026
- Google Analytics: MCP server, accessed 9 November 2026
- Claude: TikTok for Business connector, accessed 9 November 2026
- Claude: Supermetrics connector, accessed 9 November 2026
- Claude: Windsor.ai connector, accessed 9 November 2026
- Claude: Funnel connector, accessed 9 November 2026
- Claude: Contentsquare connector, accessed 9 November 2026
- Claude: Microsoft Clarity connector, accessed 9 November 2026
- Claude: Triple Whale connector, accessed 9 November 2026
- Supermetrics: Claude set-up and trial, accessed 9 November 2026
- Relevant Audience: attribution defaults in Google Ads, GA4 and Meta, 2026
- Pipeboard: Facebook Ads MCP guide (beta access reports), accessed 9 November 2026