Marketing Operations Control Centre / Agent profile

Campaign Performance Agent

Claude Cowork agent that delivers weekly campaign and portal-level performance insights with A/B test suggestions across 5 brands.

Campaign Performance Agent

Key Result: Weekly campaign & portal-level insights with A/B test suggestions Layer: Intelligence Problem: Manual, Fragmented Campaign Reporting Status: Live

Home • Agent Registry • Results • Roadmap • Campaign & Commercial Projects • Leadership & Strategy • How to Build MOCC • HubSpot Expert in South Wales


Layer: Intelligence — Performance Insight

Status: Live

Platform: Claude Cowork · Claude/HubSpot MCP Actions

Deployment: Veriforce — 5 brands

Last updated: September 2026

Version: 2.0

Business Problem

Campaign performance insights are often slow and fragmented across channels, making it difficult to understand what’s working, what’s driving pipeline, and where to optimise. Reporting tends to be manual and retrospective rather than proactive.

Solution Overview

The Campaign Performance Agent runs weekly across campaigns started in the last 50 days plus always-on campaigns, comparing performance at campaign and portal level to surface successes, concerns, engagement trends, revenue attributed and A/B test suggestions. It generates a standard insight report that highlights what changed, likely drivers and recommended next actions, and the report is emailed weekly to campaign owners, turning performance data into decision-ready insight.

Flow diagram: a weekly Claude Cowork run compares campaign and portal-level engagement and revenue through HubSpot MCP actions, then emails an insight report with A/B test suggestions

Business Value

Cost Savings

Version History

Version Date Milestone
0.1 May 2026 Design draft (inputs, outputs, cadence defined)
1.0 July 2026 Initial live version: reporting on live campaigns started in the previous 50 days
2.0 September 2026 Reworked to cover always-on campaigns and A/B test suggestions; report now shared internally with campaign owners

Design Insights

Insight outputs should be “decision first”: lead with the few changes that matter, show supporting evidence, and end with recommended actions—otherwise the agent becomes just another dashboard.


Explore Other MOCC Modules