In a nutshell: really managing paid media campaigns with AI runs through MCP, the standard that connects assistants like Claude and ChatGPT to ad platforms. The official MCPs only solve part of it: Google Ads’ only reads data, Meta’s reads and writes but only covers Meta and has limits, and LinkedIn Ads has no official MCP. Add them all up and you’re still missing LinkedIn, actions on Google, a view of the channels together, and the CRM with your business context. The LinkedScope MCP brings LinkedIn Ads, Google Ads, Meta Ads and the CRM through one connection, with a preview and your approval before every action.
Everyone wants to use AI for ads.
And most people start the same way: export a CSV from the ads manager, paste it into ChatGPT and ask it to “analyze this”.
It works. Sort of.
The AI only sees what you pasted. It doesn’t know your ICP, can’t see the CRM, doesn’t compare with Google or Meta. And when it suggests a change, you’re the one who goes and applies it by hand.
That’s not managing campaigns with AI. That’s using AI as a calculator.
In this post I’ll show how to really manage paid media campaigns with artificial intelligence: what MCP is, what the official Google, Meta and LinkedIn MCPs do today (and what they don’t), and how the LinkedScope MCP brings it all into one conversation.
What can AI do for paid media?
A lot, as long as the AI has access to the data and the tools. Using AI for paid ads isn’t just generating copy and images. With the right connection, you can ask in the chat:
- Audit: “Audit the campaigns from the last 90 days and tell me what needs attention.”
- Diagnosis: “Why did cost per lead go up this month?”
- Cross-channel comparison: “Which channel brought the most qualified leads for the same spend?”
- Creation: “Set up a campaign for our ICP with two image ads and $100 a day.”
- Cleanup: “Exclude companies that are already customers from the campaigns” or “add the negative keywords that are wasting money”.
- Pipeline: “Which target accounts warmed up and still have no open deal in the CRM?”
For that to work, the AI needs two things it doesn’t have on its own: access to the platforms and context about your business. That’s where MCP comes in.
What is MCP and why does it change campaign management with AI?
MCP (Model Context Protocol) is the open standard that lets assistants like Claude and ChatGPT use tools from other systems. Think of it as a universal plug between the AI and your platforms.
Without MCP, the AI only knows what you paste into the conversation.
With MCP, it queries data straight from the platform, picks the right tool for each question and, if the server allows it, prepares actions for you to approve.
You don’t need to code anything. Connect the MCP server to your assistant once, and that’s it: just ask.
The question is: which MCP server do you connect?
What do the official Google Ads, Meta and LinkedIn MCPs do today?
Each platform is at a different stage. Here’s where things stand in November 2026:
| Platform | Official MCP? | What it does | What it doesn’t do |
|---|---|---|---|
| Google Ads | Yes | Queries data and reports | Doesn’t create, edit or pause campaigns |
| Meta Ads | Yes | Queries, creates and manages Meta campaigns, ad sets and ads | Single-image creative (video still has limited availability); the docs don’t list lead forms or custom conversions |
| LinkedIn Ads | No | — | There’s no official MCP for LinkedIn ads |
Google Ads MCP: read-only
Google has an official MCP. It’s good for analysis: you ask, it queries the account and answers.
But Google’s own documentation makes it clear that the current version is read-only. It doesn’t create campaigns, change bids or pause anything.
And there’s a catch for non-technical folks: setup requires a Google Cloud project and OAuth credentials.
The analysis turns into a nice report. The action is still yours, by hand, inside Google Ads.
Meta MCP: reads and writes, but only Meta
Meta went further. Meta’s official MCP queries reports and creates and manages campaigns, ad sets and ads, right from the chat.
It’s real progress. But it has two limits:
- It’s partial: the creative it builds is single image (video still has limited availability), and the docs don’t list tools for lead forms or custom conversions.
- It’s Meta only: it can’t see Google, LinkedIn or your CRM.
LinkedIn Ads: no official MCP
LinkedIn doesn’t offer an official MCP for ads.
For B2B, that’s the biggest gap. The channel that targets by company and job title, and that’s usually the foundation of B2B media, is left out of the conversation with the AI.
What’s still missing even with all the official MCPs combined?
Install every official MCP that exists and you’ll still be missing:
- LinkedIn Ads, which has no MCP.
- Actions on Google Ads, which only reads.
- A view of the channels together. Each server answers only for its own channel. Combining and comparing is on you.
- The CRM. Without it, the AI optimizes clicks and leads, but doesn’t know what became an opportunity and a sale.
- Your business context. Product, offer, ICP, target accounts. You explain it all over again in every conversation.
And then there’s the operational side: one server, one login and one setup per platform.
For anyone who really manages media, especially B2B, that’s still not managing campaigns with AI. It’s having three assistants that don’t talk to each other.
How does the LinkedScope MCP manage LinkedIn Ads, Google Ads and Meta Ads in the same conversation?
With a single connection. The LinkedScope MCP brings LinkedIn Ads, Google Ads, Meta Ads and the CRM to your AI assistant, along with the briefing and ICP saved in your account.
In practice:
- Queries and actions on all three channels: create, edit, pause and activate campaigns on LinkedIn Ads, Google Ads and Meta Ads.
- Full Meta Ads: video, carousel, media upload, lead forms, custom conversions, targeting search and pre-launch checks.
- All three channels in the same answer, with the same metrics, without copying numbers from one platform to another.
- CRM in the same conversation: deals, target accounts and pipeline tied to the companies LinkedIn Ads reached.
- Saved context: briefing and ICP guide analyses, audiences, campaigns and ads. You don’t have to explain your business again.
It works in Claude, ChatGPT, Claude Code, Codex, Cursor, n8n and other clients that support remote MCP. On desktop or mobile.
Example requests
- “How much of this quarter’s pipeline came from companies LinkedIn Ads reached before the opportunity?”
- “Audit LinkedIn, Google and Meta for the last 90 days.”
- “Which target accounts warmed up in the last two weeks and still have no open deal?”
- “Exclude companies that are already customers in the CRM from the LinkedIn campaigns.”
- “Set up a LinkedIn Ads campaign for our ICP, with two ads and $100 a day. I want to review it before it goes live.”
About pipeline, one caveat that applies to any AI: influence isn’t proof that the ad caused the sale. I wrote about this in influenced pipeline.
Can you let AI touch your campaigns without risk?
Yes, as long as the AI can’t act on its own. With the LinkedScope MCP, you’re the one who decides:
- A preview before every action: you see the account, the items and the effect of the change before confirming.
- Everything starts as a draft or paused: creating a campaign or ad doesn’t start delivery. Activating is a separate confirmation.
- Permissions and logging: LinkedScope checks your account’s permissions and logs the result of every action.
- Routines keep running: the automations already set up in LinkedScope keep collecting leads, syncing audiences and sending conversions, even with the AI closed.
The AI suggests and prepares. You review and approve.
The security details are on our security page.
How do you connect the LinkedScope MCP to your assistant?
In three steps:
- Connect your accounts in LinkedScope: LinkedIn Ads, Google Ads, Meta Ads and the CRM.
- Add the LinkedScope MCP to your assistant (Claude, ChatGPT or another compatible one) and log in.
- Ask. The way you’d talk to someone on your team.
The step by step for each assistant is in the guide how to use the LinkedScope MCP. And when we release new tools, the guide how to refresh the MCP tools shows how to bring everything into your assistant.
If you’re a paid media specialist who wants to add LinkedIn to your toolkit, also check out LinkedIn Ads for paid media specialists. And to decide each channel’s role, the post on B2B paid media.
Want to manage LinkedIn Ads, Google Ads and Meta Ads in Claude or ChatGPT, with your CRM in the same conversation and your approval before every action? Meet the LinkedScope MCP and start your 15-day free trial. 😉
FAQ
How do you use artificial intelligence to manage paid media campaigns?
By connecting your AI assistant to the ad platforms through MCP. That way the AI queries data straight from the source, compares channels and prepares actions for you to approve, instead of relying on CSVs and screenshots pasted into the chat.
What is MCP?
MCP (Model Context Protocol) is the open standard that lets assistants like Claude and ChatGPT use tools from other systems. With it, the AI queries your campaigns and, if the server allows it, prepares actions.
Does Google Ads have an official MCP?
Yes, but the current version is read-only: it queries data and reports, and doesn’t create, edit or pause campaigns.
Does Meta’s official MCP create campaigns?
Yes. It queries, creates and manages Meta campaigns, ad sets and ads, but it only covers Meta and has limits: the creative is single image, and the docs don’t list lead forms or custom conversions.
Does LinkedIn Ads have an official MCP?
No. LinkedIn doesn’t offer an official MCP for ads. Through the LinkedScope MCP, you can query, create and edit LinkedIn Ads campaigns in the chat.
Is it safe to let AI touch your campaigns?
With the LinkedScope MCP, every action shows a preview and only happens after you confirm. Campaigns and ads start as drafts or paused, and every action is logged.
