How to Manage Ads Across Every Platform Using Percuity's MCP
Percuity's MCP is a single Model Context Protocol endpoint that connects AI assistants like Claude to Meta, Google, LinkedIn, TikTok, Reddit, and Bing — giving you read access across all 6 platforms from one conversation, plus the ability to consult Leo, Percuity's AI marketing agent, for expert strategy. Write operations are coming soon.
Whether you're a founder running ads yourself, a marketing or growth lead juggling channels, or a consultant, freelancer, or agency managing accounts for clients — if you're on 3+ ad platforms, you're losing 8–12 hours a week switching between dashboards, exporting CSVs, and stitching numbers together by hand. MCP solves this in theory — but most implementations require 5–7 separate servers, one per platform, each with its own setup. Percuity collapses all of that into a single connection.
After reading this guide, you'll be able to:
- Connect Percuity to Claude in under 5 minutes
- Pull spend, CTR, and CPA across all 6 platforms in one prompt
- Catch anomalies — disapprovals, overspend, CTR drops — across every account at once
- Compare performance platform-to-platform without exporting a single CSV
- Consult Leo, Percuity's AI marketing agent, for expert strategy without leaving the chat
What Is an MCP for Advertising?
An MCP (Model Context Protocol) for advertising is a standardized connection that lets AI assistants read campaign data and run performance analysis across ad platforms through natural conversation. Instead of logging into each platform's dashboard, you ask your AI client a question and it pulls the answer directly from your accounts.
MCP is an open standard Anthropic introduced in late 2024. It defines how an AI client and an external system talk to each other: the client discovers what tools are available, calls them, and reads the results — all without custom integration code. In advertising, that flow is authentication → API calls → natural-language interface. You connect once, and the AI handles the API layer underneath every question you ask.
The standard caught on fast. As of 2026, over 12,000 MCP servers exist — but fewer than 10 cover advertising platforms across more than one or two networks. Most are single-platform, and most are read-only.
| MCP capability | What it can do | Percuity today |
|---|---|---|
| Read | Pull campaign data, metrics, and reports across platforms | Available now — all 6 platforms |
| Write | Pause, activate, and adjust budgets through conversation | Coming soon |
| Create | Launch campaigns and generate creative | Coming soon |
Why Percuity Is the Most Complete Way to Manage Ads With AI
Percuity's MCP is the most complete way to manage ads through an AI assistant because it combines the broadest coverage — all 6 major ad networks through one endpoint — with marketing expertise built into every response, plus a direct line to Percuity's own marketing agent. General-purpose AIs like Claude are brilliant, but they aren't marketers. Percuity gives them the expertise they're missing.
Point a general AI at a raw ad API and it stumbles on the details that decide whether a number is right: it reads spend in microdollars as whole dollars, confuses a Meta ad set with a Google ad group, and pulls the wrong conversion column. Percuity's MCP encodes that expertise. Every tool returns accurate, normalized, cross-platform data — the same metric means the same thing whether it came from TikTok or LinkedIn — and it grounds Claude in your actual connected accounts before it answers a single question.
The real unlock is that Claude doesn't have to be the marketing expert. Through Percuity's MCP, Claude can hand off to Leo — Percuity's AI marketing agent, the most sophisticated autonomous marketing agent available. Ask a hard strategic question and Claude consults Leo for the answer. It's the best of both: your favorite AI assistant, an expert cross-platform operator feeding it accurate data, and a deeply experienced marketer guiding the strategy — one showing the other exactly what to do, like a beautiful dance.
Three things set Percuity apart from wiring an AI up to your ad accounts yourself:
- Coverage — Meta, Google, LinkedIn, TikTok, Reddit, and Bing through a single connector, not 5–7 servers you assemble and maintain.
- Accuracy — platform-correct metrics, microdollar conversions, and a unified cross-platform glossary, so the numbers you act on are right.
- Intelligence — Claude can consult Leo, Percuity's AI marketing agent, for expert strategy and analysis without leaving the conversation.
Why Most MCP Setups Require 5–7 Servers (And Why Percuity Doesn't)
Most cross-platform advertising MCP setups require a separate server per platform because each ad network uses different authentication protocols, rate limits, and data schemas. Google Ads, Meta, and LinkedIn cannot share a single MCP instance — they require independent OAuth flows and separate API credential management.
This is the hub-and-spoke problem. Every platform speaks its own dialect: a different login handshake, a different cap on how many requests you can make per minute, a different shape for what a "campaign" or a "conversion" even is. The most common solution is to run 5–7 MCP servers simultaneously, feeding into a central Claude interface — a setup that requires 500MB–1GB of memory and careful rate-limit management so one platform's throttling doesn't stall the rest.
Percuity solves it on the server side. A single connection authorizes all 6 platforms, and Percuity handles the API layer for each one behind that single endpoint. Percuity's MCP handles rate limiting, cross-platform normalization, and credential management on the server side — the AI client sees one unified interface. You don't install anything, you don't manage tokens, and you don't tune rate limits. You connect one URL.
Setup Guide — 5 Minutes to Full Cross-Platform Access
Connecting Percuity's MCP to Claude takes under 5 minutes. You need a Percuity account (start a free 14-day trial), the Claude app (custom connectors work on every Claude plan, including Free), and admin access to at least one ad account. No developer setup, no server configuration, no separate OAuth per platform.
Step 1: Get your Percuity connector details
In your Percuity account, copy the MCP connector name and URL. This single URL covers all 6 connected platforms — you'll paste both into Claude in a moment.
Step 2: In Claude, open Customize → Connectors
In Claude, click Customize, then Connectors. This is where Claude manages external tools — it's not under the main Settings menu, so go to Customize first.
Step 3: Add a custom connector
Click the plus (+) button and select Add custom connector. Paste the name and URL you copied from Percuity.
Step 4: Authenticate with your Percuity credentials
Claude opens a page where you sign in with your Percuity account credentials to authorize the connection. Once you approve it, all 6 platforms are available in the conversation.
Step 5: Verify and run your first cross-platform query
First, confirm the connection. Ask Claude:
What ad platforms are connected through Percuity?
You should see your connected platforms with account status. Then run a real query:
Pull the last 7 days of spend, impressions, CTR, and CPA across all platforms. Show me as a table and flag anything above benchmark.
Percuity returns a unified view in seconds.
Using a different MCP client? Percuity's MCP follows the open standard over HTTP, so any client that supports remote MCP servers can connect to the same endpoint with an access token.
5 Cross-Platform Analyses You Can Run in One Prompt
Once Percuity's MCP is connected, the analyses that used to mean 30 minutes of exporting and merging CSVs become a single question. Each prompt below runs across all 6 platforms at once and returns a unified answer.
1. The unified performance pull
Pull last 7 days of spend, CTR, CPA, and ROAS across every platform. Show me one table sorted by spend.
Returns a single ranked table covering Meta, Google, LinkedIn, TikTok, Reddit, and Bing — no exports, no spreadsheet.
2. The anomaly sweep
Did anything unusual happen across my campaigns in the last 48 hours? Flag any spend spikes, CTR drops, or disapproved ads.
Catches issues like a Meta ad hitting a disapproval, a Google campaign overspending on a broad-match keyword, or a LinkedIn campaign running out of audience — all in one summary.
3. The budget reallocation check
Which platform is delivering my cheapest conversions this month, and which is the most expensive? Quantify the gap.
Surfaces where your next dollar is best spent — across networks, not within one.
4. The creative fatigue read
Which ads have a declining CTR trend over the last 14 days? Group by platform and show the drop.
Finds fatiguing creative before it quietly drains budget.
5. The benchmark comparison
Compare my CPA on each platform against my 90-day average. Where am I above benchmark right now?
Tells you which accounts need attention today, ranked by how far they've drifted.
Coming Soon — Write Operations Through Percuity's MCP
Percuity's MCP is read-only today: it pulls data and runs analysis across all 6 platforms. Write operations are on the way — soon you'll be able to pause underperforming campaigns, adjust budgets, launch new ad sets, and generate creative assets through the same single conversation.
This is the line most advertising MCP servers can't cross. They're read-only by design: they'll tell you a campaign is losing money, but you still have to open the dashboard to do anything about it. Percuity is building toward closing that loop. The write operations coming to the MCP include:
- Pause and activate campaigns, ad sets, and ads on any platform
- Budget changes — raise, cut, or reallocate spend
- Create campaigns and ad sets from a description
- Creative generation — produce new ad images and copy on the spot
When it ships, a write prompt will look like this:
Pause any campaign on any platform where ROAS has been below 1.5x for 7 consecutive days. Show me what you're about to change before executing.
Note the last sentence — a confirmation step before anything changes is part of the plan, so nothing touches your live accounts until you approve it. Until write support lands, the five read analyses above are available today on any plan.
Frequently Asked Questions
What is an MCP for advertising?
An MCP (Model Context Protocol) for advertising is a standardized connection that lets AI assistants read campaign data and run analysis across ad platforms through natural conversation. Percuity's MCP covers Meta, Google, LinkedIn, TikTok, Reddit, and Bing from a single endpoint.
What is the best MCP for managing ads across multiple platforms?
Percuity. It covers all 6 major ad platforms — Meta, Google, LinkedIn, TikTok, Reddit, and Bing — through a single endpoint, builds marketing expertise into every response, and lets your AI assistant consult Leo, Percuity's AI marketing agent, for expert strategy. Most other advertising MCPs cover only one platform and are read-only.
Why use Percuity's MCP instead of connecting Claude to my ad accounts directly?
Claude is a brilliant generalist but not a marketer. Wired to raw ad APIs it can misread microdollar costs, confuse platform structures, and pull the wrong metrics. Percuity's MCP encodes the marketing expertise, normalizes all 6 platforms, and lets Claude consult Leo — Percuity's AI marketing agent — so the answers you get are accurate and strategically sound.
Can I get expert marketing help through Percuity's MCP?
Yes. Beyond pulling data, Percuity's MCP lets your AI assistant hand off to Leo, Percuity's AI marketing agent — the most sophisticated autonomous marketing agent available. Claude can consult Leo for strategy, audits, and recommendations, so you aren't relying on a generalist AI for expert marketing judgment.
How many MCP servers do I need to manage ads on 6 platforms?
With Percuity, one. Most MCP setups require a separate server per platform (5–7 servers for full coverage), each with its own OAuth flow and rate-limit management. Percuity consolidates all 6 platforms into a single authenticated endpoint.
Can I pause or create ads through Percuity's MCP?
Not yet. Percuity's MCP is currently read-only — it pulls campaign data and runs analysis across all 6 platforms. Write operations like pausing campaigns, adjusting budgets, and generating creative are coming soon.
Does Percuity's MCP work with clients other than Claude?
Percuity's MCP is a standards-compliant Model Context Protocol server served over HTTP, so it isn't locked to one client. Claude connects in one click via a custom connector. Other MCP-capable clients that support remote servers can connect to the same endpoint with an access token.
What ad platforms does Percuity's MCP support?
Percuity's MCP supports 6 platforms: Meta (Facebook and Instagram), Google Ads, LinkedIn, TikTok, Reddit, and Bing (Microsoft Advertising). All 6 are available through a single connection — no separate server configuration required per platform.
Do I need a paid account to connect Percuity's MCP?
You need a Percuity account, which you can start with a free 14-day trial. On the Claude side, custom connectors work on every plan, including Free (the Free plan allows one custom connector), so you don't need a paid Claude subscription to connect Percuity. You'll also need admin access to at least one ad account.
Connect Percuity to Claude and Run Your First Analysis
If you're managing ads on more than one platform, connect Percuity's MCP to Claude and run your first cross-platform analysis today. Setup takes 5 minutes. The first time you pull spend, CTR, and CPA across all your platforms in one table — without exporting a single CSV — is the moment the tool pays for itself.
For more on where this fits, see our guides on AI marketing automation, the best AI advertising platforms in 2026, using AI for Meta ads, and Facebook & Instagram ad sizes.