Orphex MCP

About Orphex MCP

Orphex MCP is a unified gateway that connects AI assistants to over 65 digital marketing platforms, including Google Ads, Meta, TikTok, GA4, Shopify, and AppsFlyer. It handles authentication and data integration so that tools like Claude, ChatGPT, and Gemini can analyze cross-channel performance and execute account actions from a single prompt. The system logs every change and enforces safety guardrails, requiring human approval before anything is applied to live ad accounts.

Review

Orphex MCP addresses a concrete problem for performance marketers: ad data scattered across platforms that each report their own version of the truth. Rather than blending attribution into one number, the tool surfaces each source's claim side by side and explains discrepancies. It's built for teams that want AI to do more than answer questions-they want it to act on budgets, creatives, and products within defined guardrails.

Key Features

• Multi-platform MCP integration: Connects 60+ ad platforms, analytics tools, CRMs, and e-commerce systems so AI assistants can query data across channels that individual platform MCPs cannot see.
• Action execution with approval gates: AI can reallocate budgets, pause creatives, or adjust products, but every change requires human approval. A full audit log records who requested, approved, and executed each action.
• Built-in analysis guides and custom playbooks: The tool ships with predefined analysis methods and lets teams add their own KPI definitions, targets, report templates, and brand-specific rules that the assistant follows.
• Impact Analysis module: Runs a Causal Impact model that builds a counterfactual baseline, then reports incremental lift with a 95% credible interval and significance verdict, using data already in connected platforms.
• Workspace isolation for agencies: Each client gets a separate workspace with its own platform connections, role-based access (Viewer, Viewer Plus, Editor, Admin), and brand-specific memory, guardrails, and scorecards.

Pricing and Value

Orphex MCP offers a free tier that does not require a credit card. Paid plans are available, and the launch period includes 30% off the first month. Specific plan tiers, feature limits, and post-discount pricing have not been detailed in the available reference content. The tool currently serves over 100 brands managing a combined $100M+ in monthly ad spend.

Pros

• Reads attribution data from each platform separately instead of forcing a single blended number, making discrepancies visible.
• Impact Analysis module handles all data preparation-time series, pre/post windows, control signals-without requiring user exports.
• Workspace structure with per-brand memory and guardrails means multi-client agencies can enforce different rules for each account.
• Audit log merges Orphex's change history with each ad platform's own log, creating one timeline of edits regardless of where they were made.
• Conversion activation sends offline CRM outcomes (won deals, revenue) back to ad platforms as server-side conversions for optimization.

Cons

• Native CRM support is limited to HubSpot, Insider, and D-engage; Pipedrive and other CRMs require workarounds like Google Sheets or BigQuery sync until native integrations ship.
• The tool is not well suited for teams that want fully autonomous AI optimization without approval steps-every change must be reviewed and approved by a human.
• Non-e-commerce and non-app-install funnels rely on the conversion activation feature, which is listed as early access and may not yet cover all CRM-stage outcomes natively.

Orphex MCP fits agencies and in-house teams running multi-channel paid media who need cross-platform context for their AI assistants and can work within a human-approval workflow. It's less practical for solo operators who want hands-off automation or for organizations whose CRM isn't among the three natively supported platforms. The tool's strength lies in making platform-level attribution differences explicit and turning AI from a query engine into a supervised action layer.