Marketing Blueprint
Safety checkedAI Campaign Factory: Meta Ads, Google Ads & WhatsApp with Claude + Clint
Avoid optimizing for clicks that never become pipeline: build reviewed Meta, Google, and WhatsApp campaigns around CRM audience segments and qualified-lead outcomes. Human approval gates launches, while daily lead-quality checks surface proposed adjustments.
Install Prompt
Copy this prompt and paste it into your AI agent to install.
What problem does this solve?
Turns a campaign brief and first-party performance data into reviewed campaigns across Meta Ads, Google Ads, and WhatsApp. It supports B2B SaaS marketers running demand generation who need to target audiences, create channel-specific copy, and evaluate results using qualified leads and other CRM outcomes rather than clicks alone.
How does it work?
- Give Claude the campaign objective, offer, channel mix, budget, and dates. Claude presents ClickHouse SQL queries for approval before running them.
- Use the resulting audience segments and multi-touch performance data to draft channel-specific creative and copy. The copy reviewer scores and rewrites each piece until it reaches at least 90/100, then a marketer approves the plan.
- Create WhatsApp templates and broadcasts in Clint, and build Meta campaigns through MCP in paused status. Prepare Google Ads changes for human validation. A human chooses send times and budgets and activates paid campaigns.
- Send CRM outcomes such as qualified leads, meetings, and sales back to Meta and Google. Run a scheduled daily lead-quality check to propose campaign adjustments. Output: reviewed multichannel campaigns, audience data, and daily lead-quality recommendations.
What's the biggest win?
One marketer goes from brief to live multichannel campaigns (Meta Ads, Google Ads and WhatsApp broadcasts) in hours instead of days. Audiences come from clean first-party CRM data (no overlap, opt-outs removed), copy only ships after scoring 90+/100, and platforms optimize for qualified leads and meetings sent back from the CRM instead of clicks. Daily automated lead-quality checks turn yesterday's results into today's action plan.
What's required to run this?
- Claude Cowork or Claude Code with Skills enabled: orchestrates the workflow and runs scheduled daily routines.
- Clint CRM with Clint MCP: manages WhatsApp templates, contacts, tags, broadcasts, and automations.
- ClickHouse MCP with read-only access: supplies CRM data for audience and performance analysis; the workflow presents SQL for approval before querying production data.
- Official Meta Ads MCP: creates campaigns, ad sets, ads, audiences, and insights. Restrict access to one ad account; campaigns remain paused for human review.
- Google Ads: used for search campaigns and offline conversions; validate changes before execution.
- Claude in Chrome: fallback for Ads Manager settings that the Meta MCP cannot create.
- Modular Skills:
data-base-builder,creative-director,copy-reviewer,whatsapp-template,import-base,schedule-broadcast,full-broadcast,meta-campaign-builder, andlead-quality-check. - Copy review threshold: each piece must score at least 90/100 before approval.
What are the constraints?
Production-data queries require approval before execution, except pre-approved read-only scheduled routines. Paid campaigns must be created paused and require human activation. Humans choose send times and budgets, and Google Ads changes require validation. WhatsApp audiences must exclude opt-outs and blacklists, and every template must include a block-messages button. The workflow should duplicate campaigns rather than modify live originals. Cold click-to-WhatsApp audiences and lookalikes attracted curious users in the reported test; warm audiences performed better for qualified pipeline.
Tools in this Blueprint
About This Blueprint
- Industry
- Computer Software
More Blueprints to explore
WhatsApp Appointment Booking for Clinics: AI Agent That Books and Confirms Visits
WhatsApp appointment booking for clinics: an AI agent books patients 24/7 on the right professional's Google Calendar, an automation confirms every visit before the appointment, and reschedules, cancellations and missed visits are handled in the conversation and tracked in the CRM.
AI Lead Qualification on WhatsApp: Score Leads and Send the Hot Ones to Sales
AI lead qualification on WhatsApp: an AI agent asks up to four questions, scores fit and intent with a written rubric, and sends hot leads to a sales rep with an owner and the reasoning behind the score, while warm and cold leads follow their own path in the CRM.
WhatsApp Chatbot Not Closing Sales? Diagnose It and Upgrade to an AI Agent
WhatsApp chatbot not closing sales? Diagnose it with 20 lost conversations and 10 live tests, then decide between fixing the bot and upgrading to an AI agent that understands free text and audio, books the next step, hands off and follows up.