run it from your own AI
MCP marketing: give your AI the whole marketing stack.
The Model Context Protocol connects AI agents to your marketing tools. Cogny is one MCP server with 72 integrations and 1,378 tools behind it, so Claude, ChatGPT, Cursor or Codex can read live data and make changes. Cogny keeps the ledger: every write logged, launches measured automatically.
or from your own AI
claude mcp add --transport http cogny https://app.cogny.com/mcpthe model context protocol, explained for marketers
MCP marketing is using the Model Context Protocol to connect AI agents directly to marketing platforms and data. MCP is an open standard introduced by Anthropic: a single interface that lets any AI client discover the tools available, read data through them and take actions with them, without a custom integration per tool.
Before MCP, connecting AI to your marketing meant pasting CSVs into a chat or building API integrations, each with its own authentication, format and upkeep. With MCP, your AI client sees tools such as “get campaign insights” or “update ad set budget” and calls them like a function.
Cogny is the marketing MCP server. Add it once, authorise in the browser, and connect the tools you use: Google Ads, Meta Ads, GA4, Search Console, Tag Manager, LinkedIn, TikTok, Klaviyo, Mailchimp, Shopify, WordPress, Webflow and more, 72 in all. Each connection uses the platform's own sign-in, and you can disconnect any of them at any time.
What MCP alone doesn't give you is memory. Cogny adds it. Every write call through Cogny is logged with the tool, the time, the result and who authorised it, launches your AI makes, such as new campaigns and ads, become tickets automatically; for other changes your AI offers to register them. Cogny measures each ticket three days later for ads and seven for content and SEO, then closes it with a verdict. On Cloud, your AI can also work an assigned ticket end to end.
Your AI does the work. Cogny keeps the ledger.
from connection to measured change in four steps
- 01
Add Cogny
One line in Claude Code or Codex, a connector in Claude.ai or ChatGPT, or .mcp.json in Cursor. Then authorise in the browser.
- 02
Connect your stack
Sign in to Google Ads, Meta, GA4 and the rest with their own OAuth.
- 03
Ask and act
Your AI reads live data and proposes changes. Its approval prompt is where you say yes.
- 04
Verdict
Cogny logs the write. Launches become tickets automatically; other changes your AI offers to register. Then a verdict.
changes made from your own AI, on the record
Registered changes land on the same ledger, whichever client made them.
Example rows, not customer data.
why cogny, and not a pile of separate MCP servers
One server, the whole stack
72 integrations and 1,378 tools behind one connection. Each tool is marked read, write or destructive.
Closed-loop measurement
Launches become tickets automatically and your AI offers to register other changes. Each is measured and closed as worked, partial, didn't work or inconclusive.
A log of every write
Tool, time, result and who authorised it. Blocked tools are refused.
Cogny's context, in your AI
Your brand context, measurement plan and past verdicts come with the tools.
where MCP marketing earns its keep
Paid media across platforms
Your AI reads Google Ads, Meta and LinkedIn in one pass and proposes the change worth making next. The research that took an analyst a day takes a conversation.
SEO and AI search
Search Console, Bing Webmaster Tools and an AI visibility check in the same session, so your AI can see which pages win clicks and which get cited by AI assistants.
Tracking and funnels
With Tag Manager, GA4 and the ad platforms connected, your AI can find a broken conversion tag or a funnel gap and fix it, with the change measured like any other.
Cross-platform reporting
One question reads every connected platform instead of five dashboards and an export. Ask once, and a finished chat can become a Loop that reruns weekly on Solo, Cogny AI and Cogny Pro; on Cloud, scheduled reports run on their own and end in tickets.
Anomalies, the next morning
Paid plans get a daily update on the start page's KPI graphs, so a CTR drop, a CPM spike or a conversion tag that stopped firing shows up with the data behind it.
Audiences and creative
Your AI reads audience and ad performance across Meta, Google and TikTok in one pass, finds fatigue and overlap, and proposes the one change worth making, measured three days later.
what it doesn't do
- It doesn't add a second approval to your AI's. Over MCP, your client's approval prompt is the gate; Cogny logs every write.
- It doesn't hide your credentials in a prompt. Integrations sign in with each platform's own OAuth.
- It doesn't meter Solo by the call. Solo is unmetered under fair use; Free has 15 MCP calls a month.
the record.
Published case studies, each one linked.
common questions about MCP marketing and cogny
What is MCP marketing?
MCP marketing refers to using the Model Context Protocol, an open standard introduced by Anthropic, to connect AI agents directly to your marketing tools and data. Instead of exporting CSVs or switching between dashboards, MCP lets an AI agent read live data from Google Ads, Meta, GA4 and Search Console, then act on it: pausing campaigns, adjusting budgets, updating audiences. Cogny is a marketing MCP server built on this protocol.
How is Cogny different from other AI marketing tools?
Most AI marketing tools give you a chatbot that works with whatever data you paste in. Cogny connects your AI to your actual stack over MCP, so it reads live numbers and can make approved changes in the platforms. Launches your AI makes become tickets automatically, other changes it offers to register, and Cogny measures each in your own data and closes it with a verdict: analyse, decide, act, measure.
What marketing tools does Cogny connect to via MCP?
72 integrations, including Google Ads, Meta Ads, Google Analytics 4, Google Search Console, Google Tag Manager, LinkedIn Ads, TikTok Ads, Microsoft Advertising, Klaviyo, Mailchimp, Shopify, WooCommerce, WordPress and Webflow. Each connection signs in with the platform's own OAuth or API key, and you can disconnect any tool at any time. The full list, with every tool, is at /integrations/tools.
What does “AI takes action” mean? Is it safe?
When your AI proposes a change, say pausing an ad set, it calls the Cogny tool for it. Over MCP, your AI client asks you before that write runs; that prompt is the approval step, and Cogny does not hold it for a second one. In the Cogny app, write tools ask first by default, per tool. Tools set to blocked are refused. Every write is logged with the tool, the time, the result and who authorised it. Launches are measured as tickets automatically; other changes are measured once registered.
How much does Cogny cost?
Free is $0 with 15 MCP calls a month and 50 one-time chat credits, no card. Solo is $9/month and unmetered over MCP under fair use. Cogny AI ($18) and Cogny Pro ($90) add monthly credits for in-app chat. Cloud is $499/month with 5,000 credits, team access, per-member MCP controls, scheduled reports and priority support. There is no free trial on paid plans; Free is the way to try it.
Which marketing MCP servers can you trust?
Look for OAuth sign-in per platform, a published list of every tool marked read, write or destructive, a log of every write with who authorised it, and the ability to block any tool. Cogny publishes all of its tools at cogny.com/integrations/tools. From your own AI, your client asks before each write; Cogny logs it and turns launches into measured tickets.
Which AI clients work with Cogny over MCP?
Claude Code and Codex with one install line, Claude.ai and ChatGPT as a connector, and Cursor through .mcp.json. All of them reach the same tools and the same ledger, so a change made from one client is measured the same way as a change made from another, or from the Cogny app.
Can I connect Cogny without IT help?
Yes. Adding Cogny to your AI client is one line or one connector, then a browser sign-in. Each integration is connected by signing in to that platform, with no API keys to manage for OAuth tools and no custom code.
what it costs.
- Free $015 MCP calls a month and 50 one-time chat credits, no card
- Solo $9unmetered in your own AI, fair use
- Cogny AI $18150 credits
- Cogny Pro $901,500 credits
- Cloud $4995,000 credits, for teams
Per month. Solo, Cogny AI and Cogny Pro have one seat each. Every change is recorded on every plan; Solo measures 3 a month, Cogny AI and Pro measure every one and add monthly credits. Teams need Cloud. Compare plans →
your AI, with a memory.