Claude marketing · Claude for marketing
Claude marketing where every change gets a verdict.
Claude is a strong analyst. Cogny is the harness that turns it into a marketing system: your tools connected over MCP, every change approved and logged, and a verdict three to seven days later.
72 integrations, 1,380 tools, the same on every paid plan. Start on Solo at $9/mo with your own Claude; move a team to Cloud at $499/mo for scheduled reports and ticket generation.
raw Claude
a brilliant analyst with no hands and no memory.
- 01
Pasted data
Without tools, Claude sees what you export and paste: a sample, already out of date. - 02
Prose, not changes
It writes good recommendations. Someone still has to make them in four ad managers. - 03
No record
Next week it has forgotten what it suggested, and nobody checked whether it worked. - 04
No schedule
It runs when you remember to ask, not when the account needs it.
how cogny does it
the harness around Claude.
- 01
experiment across your whole stack.
Claude changes a bid in Google Ads and reads the result in GA4 or your CRM. 72 integrations, read and write. - 02
from raw data to a shipped change.
A question, one next step, your yes in Claude, the change made in the tool. Then a ticket and a verdict. - 03
cogny keeps the ledger.
Every write logged with who authorised it. Launches from Claude become measured tickets automatically; other changes Claude offers to register. A finished chat can become a Loop that runs again on a schedule.
why Claude for marketing
Marketing is mostly numbers in rows. Claude reads schemas, joins tables and says what it can't tell from the data. Read the comparison
Built for analytical reasoning
It explains relationships across tables, which is most of the work in a marketing account.Calibrated uncertainty
It separates what it knows from what it suspects: the difference between useful and dangerous in production.MCP-native
Anthropic introduced the Model Context Protocol; the marketing MCP ecosystem grew up around Claude first.Your assistant, your choice
The same Cogny tools work from ChatGPT, Cursor and Codex, so nothing is locked to one model.
the AI marketing harness, layer by layer
The model is half the story. Read the harness piece
- 01
72 live MCP integrations
Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads, GA4, Search Console, Klaviyo, Mailchimp, HubSpot, Symplify, Shopify and more: connected once, authenticated, maintained. - 02
Context that persists
A workspace memory of your company, offers, brand voice and competitors, built from your website at signup and read in every conversation. - 03
Loops and scheduled reports
On Solo, Cogny AI and Cogny Pro a finished chat becomes a Loop that runs on a schedule. Cloud adds scheduled reports from 70+ templates. - 04
Measured tickets
Every shipped change becomes a ticket with the expected outcome written down, so a recommendation is a hypothesis, not a vibe. - 05
Approval and a write log
Claude asks before a write; Cogny logs tool, time, result and who authorised it. Blocked tools are refused. - 06
The Truth Ledger
Each ticket closes as worked, partial, didn't work or inconclusive. Misses stay on the record so they aren't repeated.
Claude marketing use cases
Paid media audits
Google Ads waste, Meta creative fatigue, LinkedIn audience overlap, cross-channel budget, search-term mining.AI SEO + GEO
Search Console deep dives, AI citation tracking, content gaps, keyword cannibalisation, GEO content briefs.Email & lifecycle
Segment engagement decay, flow performance, subject-line patterns, list health.Revenue & cohorts
Cohort revenue, B2B pipeline signals, CRO lifts, churn, attribution reconciliation.
raw Claude vs. free skills vs. the cogny harness
| Capability | Raw Claude | Claude + free skills | Cogny harness |
|---|---|---|---|
| Data access | Pasted text | You build the MCP | 72 integrations included |
| Changes in the tools | Copy and paste | If you wire it | Read and write tools, approved in Claude |
| Schedule | None | None | Loops; scheduled reports on Cloud |
| Output shape | Prose | Better prose | A ticket per change |
| Audit trail | None | None | Every write logged |
| Outcome measurement | None | None | A verdict after 3–7 days |
| Monthly cost | Your Claude plan | Your Claude plan + engineering time | Free $0 · Solo $9 · Cloud $499 |
further reading
- AI marketing harness vs. raw ClaudeWhy a chat window can't replace the harness layer.
- Claude for marketing: a practical guideWhat Claude is good at for paid media, SEO, GEO, email and content.
- Falsifiable marketing experimentsWhy AI tickets beat AI suggestions.
- AI marketing is the new SEOOne strategy for Google and the AI engines.
Every integration and tool: integrations directory.
the record.
works with.
questions.
What is Claude marketing?
Claude marketing means using Anthropic's Claude for production marketing work: analysing campaigns, finding what to fix, optimising SEO and GEO, reading email performance, and making approved changes in live ad accounts. With Cogny, Claude reaches your marketing tools over MCP, and the launches it makes become measured tickets automatically, and it offers to register other changes so they are measured too.
Why Claude for marketing?
Claude is good at the parts marketing needs: reasoning over tables of numbers, saying what it can't tell from the data, and calling tools over MCP, the protocol Anthropic introduced. Cogny also works from ChatGPT, Cursor and Codex, so the choice of assistant stays yours.
What is an AI marketing harness?
An AI marketing harness is the production layer around a model: live tool access over MCP, an approval step, a measured ticket for each launch, a schedule, and measurement of the outcome. Raw Claude is a calculator. The harness is what turns it into a system that remembers what worked. Cogny is the harness for Claude marketing.
How many integrations and templates does Cogny ship?
72 integrations with 1,380 tools, from Google Ads, Meta Ads, LinkedIn Ads, TikTok Ads, GA4, Search Console and Klaviyo to Shopify, HubSpot and WordPress. Every paid plan can connect every integration; the tools are the same on Solo as on Cloud. Cloud adds scheduled reports from 70+ templates covering paid media, SEO, GEO, email, revenue and B2B pipeline.
What makes Cogny different from a free skills install?
Free skills are reusable prompt scaffolds, useful if you are building your own harness. Cogny is the harness already built: the integrations maintained, a log of every write, each shipped change turned into a ticket and measured, and Loops that repeat a finished chat on a schedule. On Cloud, scheduled reports and ticket generation run across every channel without you.
Can Claude execute changes against my ad accounts?
Yes. Over MCP, Claude asks you before it runs a tool that changes something; that prompt is the approval. Cogny logs each write with who authorised it, and a tool you set to blocked in Cogny is refused. In the Cogny app, write tools ask first by default. Launches Claude makes become measured tickets automatically, and it offers to register other changes; each ticket is measured three days later for ads and seven for content and SEO.
What does Claude marketing cost?
Free is $0 with 15 MCP calls a month. Solo is $9/month with unmetered MCP calls from your own Claude, under fair use; your Claude subscription pays for the thinking. Cogny AI ($18) and Cogny Pro ($90) add monthly credits for Cogny's own in-app chat. Cloud is $499/month for teams, with scheduled reports, ticket generation, the BigQuery warehouse and priority support. No free trial on paid plans; Free is the way to try it.
How does Claude marketing handle SEO and GEO?
Claude reads Search Console, GA4 (including referrals from ChatGPT, Perplexity and Gemini), Bing Webmaster Tools and AI visibility checks in one conversation, finds content opportunities for both Google and the AI engines, and turns the one you approve into a ticket that is measured seven days later.
How long until I see results from Claude marketing?
The first analysis comes back in the same conversation you connect in. Each change gets its verdict on a fixed clock: three days after an ads change, seven days after a content or SEO change. Ranking and AI-citation gains compound over months; the published case studies show the time windows that were measured.
Is my data used to train Claude?
When you drive Cogny from your own Claude, your prompts go to Anthropic under your own account's terms. Cogny's MCP server does not store or cache query results; data is fetched from the connected platform on demand. Cogny's subprocessors for its in-app AI do not train on customer content submitted via their APIs. The details are in the DPA at /terms/dpa.
Claude. connected. graded.
Copy the line into Claude Code, or add Cogny as a connector in Claude.ai.
claude mcp add --transport http cogny https://app.cogny.com/mcpFree $0, no card · Solo $9 · Cogny AI $18, 150 credits · Cogny Pro $90, 1,500 credits · Cloud $499, 5,000 credits, for teams. Compare plans